🔧 WIP: Enhance Bedrock endpoint configuration with user-provided credentials. (Still needs to implement user_provided bearer token support, but the UI is there for it)

- Added support for user-provided AWS credentials (Access Key ID, Secret Access Key, Session Token, Bearer Token) in the Bedrock endpoint configuration. - Localized new strings for Bedrock configuration in translation files.
🪵 refactor: Dynamic getLogDirectory utility for Loggers (#8686 )
2025-07-26 18:14:02 -07:00 · 2025-07-26 20:11:20 -04:00 · 2025-07-25 14:51:42 -04:00 · 2025-07-25 14:06:16 -04:00 · 2025-07-25 13:45:22 -04:00 · 2025-07-25 12:33:05 -04:00
1261 changed files with 115522 additions and 33083 deletions
--- a/.env.example
+++ b/.env.example
@@ -20,8 +20,8 @@ DOMAIN_CLIENT=http://localhost:3080
 DOMAIN_SERVER=http://localhost:3080

 NO_INDEX=true
-# Use the address that is at most n number of hops away from the Express application. 
-# req.socket.remoteAddress is the first hop, and the rest are looked for in the X-Forwarded-For header from right to left. 
+# Use the address that is at most n number of hops away from the Express application.
+# req.socket.remoteAddress is the first hop, and the rest are looked for in the X-Forwarded-For header from right to left.
 # A value of 0 means that the first untrusted address would be req.socket.remoteAddress, i.e. there is no reverse proxy.
 # Defaulted to 1.
 TRUST_PROXY=1
@@ -58,7 +58,7 @@ DEBUG_CONSOLE=false
 #                     Endpoints                     #
 #===================================================#

-# ENDPOINTS=openAI,assistants,azureOpenAI,google,gptPlugins,anthropic
+# ENDPOINTS=openAI,assistants,azureOpenAI,google,anthropic

 PROXY=

@@ -88,7 +88,7 @@ PROXY=
 #============#

 ANTHROPIC_API_KEY=user_provided
-# ANTHROPIC_MODELS=claude-3-5-haiku-20241022,claude-3-5-sonnet-20241022,claude-3-5-sonnet-latest,claude-3-5-sonnet-20240620,claude-3-opus-20240229,claude-3-sonnet-20240229,claude-3-haiku-20240307,claude-2.1,claude-2,claude-1.2,claude-1,claude-1-100k,claude-instant-1,claude-instant-1-100k
+# ANTHROPIC_MODELS=claude-opus-4-20250514,claude-sonnet-4-20250514,claude-3-7-sonnet-20250219,claude-3-5-sonnet-20241022,claude-3-5-haiku-20241022,claude-3-opus-20240229,claude-3-sonnet-20240229,claude-3-haiku-20240307
 # ANTHROPIC_REVERSE_PROXY=

 #============#
@@ -142,12 +142,12 @@ GOOGLE_KEY=user_provided
 # GOOGLE_AUTH_HEADER=true

 # Gemini API (AI Studio)
-# GOOGLE_MODELS=gemini-2.0-flash-exp,gemini-2.0-flash-thinking-exp-1219,gemini-exp-1121,gemini-exp-1114,gemini-1.5-flash-latest,gemini-1.0-pro,gemini-1.0-pro-001,gemini-1.0-pro-latest,gemini-1.0-pro-vision-latest,gemini-1.5-pro-latest,gemini-pro,gemini-pro-vision
+# GOOGLE_MODELS=gemini-2.5-pro,gemini-2.5-flash,gemini-2.5-flash-lite-preview-06-17,gemini-2.0-flash,gemini-2.0-flash-lite

 # Vertex AI
-# GOOGLE_MODELS=gemini-1.5-flash-preview-0514,gemini-1.5-pro-preview-0514,gemini-1.0-pro-vision-001,gemini-1.0-pro-002,gemini-1.0-pro-001,gemini-pro-vision,gemini-1.0-pro
+# GOOGLE_MODELS=gemini-2.5-pro,gemini-2.5-flash,gemini-2.5-flash-lite-preview-06-17,gemini-2.0-flash-001,gemini-2.0-flash-lite-001

-# GOOGLE_TITLE_MODEL=gemini-pro
+# GOOGLE_TITLE_MODEL=gemini-2.0-flash-lite-001

 # GOOGLE_LOC=us-central1

@@ -175,7 +175,7 @@ GOOGLE_KEY=user_provided
 #============#

 OPENAI_API_KEY=user_provided
-# OPENAI_MODELS=o1,o1-mini,o1-preview,gpt-4o,chatgpt-4o-latest,gpt-4o-mini,gpt-3.5-turbo-0125,gpt-3.5-turbo-0301,gpt-3.5-turbo,gpt-4,gpt-4-0613,gpt-4-vision-preview,gpt-3.5-turbo-0613,gpt-3.5-turbo-16k-0613,gpt-4-0125-preview,gpt-4-turbo-preview,gpt-4-1106-preview,gpt-3.5-turbo-1106,gpt-3.5-turbo-instruct,gpt-3.5-turbo-instruct-0914,gpt-3.5-turbo-16k
+# OPENAI_MODELS=o1,o1-mini,o1-preview,gpt-4o,gpt-4.5-preview,chatgpt-4o-latest,gpt-4o-mini,gpt-3.5-turbo-0125,gpt-3.5-turbo-0301,gpt-3.5-turbo,gpt-4,gpt-4-0613,gpt-4-vision-preview,gpt-3.5-turbo-0613,gpt-3.5-turbo-16k-0613,gpt-4-0125-preview,gpt-4-turbo-preview,gpt-4-1106-preview,gpt-3.5-turbo-1106,gpt-3.5-turbo-instruct,gpt-3.5-turbo-instruct-0914,gpt-3.5-turbo-16k

 DEBUG_OPENAI=false

@@ -209,12 +209,6 @@ ASSISTANTS_API_KEY=user_provided
 # More info, including how to enable use of Assistants with Azure here:
 # https://www.librechat.ai/docs/configuration/librechat_yaml/ai_endpoints/azure#using-assistants-with-azure

-#============#
-# OpenRouter #
-#============#
-# !!!Warning: Use the variable above instead of this one. Using this one will override the OpenAI endpoint
-# OPENROUTER_API_KEY=
-
 #============#
 # Plugins    #
 #============#
@@ -237,6 +231,14 @@ AZURE_AI_SEARCH_SEARCH_OPTION_QUERY_TYPE=
 AZURE_AI_SEARCH_SEARCH_OPTION_TOP=
 AZURE_AI_SEARCH_SEARCH_OPTION_SELECT=

+# OpenAI Image Tools Customization
+#----------------
+# IMAGE_GEN_OAI_DESCRIPTION_WITH_FILES=Custom description for image generation tool when files are present
+# IMAGE_GEN_OAI_DESCRIPTION_NO_FILES=Custom description for image generation tool when no files are present
+# IMAGE_EDIT_OAI_DESCRIPTION=Custom description for image editing tool
+# IMAGE_GEN_OAI_PROMPT_DESCRIPTION=Custom prompt description for image generation tool
+# IMAGE_EDIT_OAI_PROMPT_DESCRIPTION=Custom prompt description for image editing tool
+
 # DALL·E
 #----------------
 # DALLE_API_KEY=
@@ -254,6 +256,13 @@ AZURE_AI_SEARCH_SEARCH_OPTION_SELECT=
 # DALLE3_AZURE_API_VERSION=
 # DALLE2_AZURE_API_VERSION=

+# Flux
+#-----------------
+FLUX_API_BASE_URL=https://api.us1.bfl.ai
+# FLUX_API_BASE_URL = 'https://api.bfl.ml';
+
+# Get your API key at https://api.us1.bfl.ai/auth/profile
+# FLUX_API_KEY=

 # Google
 #-----------------
@@ -340,6 +349,11 @@ REGISTRATION_VIOLATION_SCORE=1
 CONCURRENT_VIOLATION_SCORE=1
 MESSAGE_VIOLATION_SCORE=1
 NON_BROWSER_VIOLATION_SCORE=20
+TTS_VIOLATION_SCORE=0
+STT_VIOLATION_SCORE=0
+FORK_VIOLATION_SCORE=0
+IMPORT_VIOLATION_SCORE=0
+FILE_UPLOAD_VIOLATION_SCORE=0

 LOGIN_MAX=7
 LOGIN_WINDOW=5
@@ -363,7 +377,7 @@ ILLEGAL_MODEL_REQ_SCORE=5
 # Balance                #
 #========================#

-CHECK_BALANCE=false
+# CHECK_BALANCE=false
 # START_BALANCE=20000 # note: the number of tokens that will be credited after registration.

 #========================#
@@ -431,15 +445,60 @@ OPENID_NAME_CLAIM=

 OPENID_BUTTON_LABEL=
 OPENID_IMAGE_URL=
+# Set to true to automatically redirect to the OpenID provider when a user visits the login page
+# This will bypass the login form completely for users, only use this if OpenID is your only authentication method
+OPENID_AUTO_REDIRECT=false
+# Set to true to use PKCE (Proof Key for Code Exchange) for OpenID authentication
+OPENID_USE_PKCE=false
+#Set to true to reuse openid tokens for authentication management instead of using the mongodb session and the custom refresh token.
+OPENID_REUSE_TOKENS=
+#By default, signing key verification results are cached in order to prevent excessive HTTP requests to the JWKS endpoint.
+#If a signing key matching the kid is found, this will be cached and the next time this kid is requested the signing key will be served from the cache.
+#Default is true.
+OPENID_JWKS_URL_CACHE_ENABLED=
+OPENID_JWKS_URL_CACHE_TIME= # 600000 ms eq to 10 minutes leave empty to disable caching
+#Set to true to trigger token exchange flow to acquire access token for the userinfo endpoint.
+OPENID_ON_BEHALF_FLOW_FOR_USERINFO_REQUIRED=
+OPENID_ON_BEHALF_FLOW_USERINFO_SCOPE="user.read" # example for Scope Needed for Microsoft Graph API
+# Set to true to use the OpenID Connect end session endpoint for logout
+OPENID_USE_END_SESSION_ENDPOINT=
+
+
+# SAML
+# Note: If OpenID is enabled, SAML authentication will be automatically disabled.
+SAML_ENTRY_POINT=
+SAML_ISSUER=
+SAML_CERT=
+SAML_CALLBACK_URL=/oauth/saml/callback
+SAML_SESSION_SECRET=
+
+# Attribute mappings (optional)
+SAML_EMAIL_CLAIM=
+SAML_USERNAME_CLAIM=
+SAML_GIVEN_NAME_CLAIM=
+SAML_FAMILY_NAME_CLAIM=
+SAML_PICTURE_CLAIM=
+SAML_NAME_CLAIM=
+
+# Logint buttion settings (optional)
+SAML_BUTTON_LABEL=
+SAML_IMAGE_URL=
+
+# Whether the SAML Response should be signed.
+# - If "true", the entire `SAML Response` will be signed.
+# - If "false" or unset, only the `SAML Assertion` will be signed (default behavior).
+# SAML_USE_AUTHN_RESPONSE_SIGNED=
+

 # LDAP
 LDAP_URL=
 LDAP_BIND_DN=
 LDAP_BIND_CREDENTIALS=
 LDAP_USER_SEARCH_BASE=
-LDAP_SEARCH_FILTER=mail={{username}}
+#LDAP_SEARCH_FILTER="mail="
 LDAP_CA_CERT_PATH=
 # LDAP_TLS_REJECT_UNAUTHORIZED=
+# LDAP_STARTTLS=
 # LDAP_LOGIN_USES_USERNAME=true
 # LDAP_ID=
 # LDAP_USERNAME=
@@ -461,6 +520,18 @@ EMAIL_PASSWORD=
 EMAIL_FROM_NAME=
 EMAIL_FROM=noreply@librechat.ai

+#========================#
+#      Mailgun API       #
+#========================#
+
+# MAILGUN_API_KEY=your-mailgun-api-key
+# MAILGUN_DOMAIN=mg.yourdomain.com
+# EMAIL_FROM=noreply@yourdomain.com
+# EMAIL_FROM_NAME="LibreChat"
+
+# # Optional: For EU region
+# MAILGUN_HOST=https://api.eu.mailgun.net
+
 #========================#
 # Firebase CDN           #
 #========================#
@@ -472,6 +543,24 @@ FIREBASE_STORAGE_BUCKET=
 FIREBASE_MESSAGING_SENDER_ID=
 FIREBASE_APP_ID=

+#========================#
+# S3 AWS Bucket          #
+#========================#
+
+AWS_ENDPOINT_URL=
+AWS_ACCESS_KEY_ID=
+AWS_SECRET_ACCESS_KEY=
+AWS_REGION=
+AWS_BUCKET_NAME=
+
+#========================#
+# Azure Blob Storage     #
+#========================#
+
+AZURE_STORAGE_CONNECTION_STRING=
+AZURE_STORAGE_PUBLIC_ACCESS=false
+AZURE_CONTAINER_NAME=files
+
 #========================#
 # Shared Links           #
 #========================#
@@ -491,6 +580,10 @@ ALLOW_SHARED_LINKS_PUBLIC=true
 # If you have another service in front of your LibreChat doing compression, disable express based compression here
 # DISABLE_COMPRESSION=true

+# If you have gzipped version of uploaded image images in the same folder, this will enable gzip scan and serving of these images
+# Note: The images folder will be scanned on startup and a ma kept in memory. Be careful for large number of images.
+# ENABLE_IMAGE_OUTPUT_GZIP_SCAN=true
+
 #===================================================#
 #                        UI                         #
 #===================================================#
@@ -508,11 +601,40 @@ HELP_AND_FAQ_URL=https://librechat.ai
 # REDIS Options #
 #===============#

-# REDIS_URI=10.10.10.10:6379
+# Enable Redis for caching and session storage
 # USE_REDIS=true

-# USE_REDIS_CLUSTER=true
-# REDIS_CA=/path/to/ca.crt
+# Single Redis instance
+# REDIS_URI=redis://127.0.0.1:6379
+
+# Redis cluster (multiple nodes)
+# REDIS_URI=redis://127.0.0.1:7001,redis://127.0.0.1:7002,redis://127.0.0.1:7003
+
+# Redis with TLS/SSL encryption and CA certificate
+# REDIS_URI=rediss://127.0.0.1:6380
+# REDIS_CA=/path/to/ca-cert.pem
+
+# Redis authentication (if required)
+# REDIS_USERNAME=your_redis_username
+# REDIS_PASSWORD=your_redis_password
+
+# Redis key prefix configuration
+# Use environment variable name for dynamic prefix (recommended for cloud deployments)
+# REDIS_KEY_PREFIX_VAR=K_REVISION
+# Or use static prefix directly
+# REDIS_KEY_PREFIX=librechat
+
+# Redis connection limits
+# REDIS_MAX_LISTENERS=40
+
+# Redis ping interval in seconds (0 = disabled, >0 = enabled)
+# When set to a positive integer, Redis clients will ping the server at this interval to keep connections alive
+# When unset or 0, no pinging is performed (recommended for most use cases)
+# REDIS_PING_INTERVAL=300
+
+# Force specific cache namespaces to use in-memory storage even when Redis is enabled
+# Comma-separated list of CacheKeys (e.g., STATIC_CONFIG,ROLES,MESSAGES)
+# FORCED_IN_MEMORY_CACHE_NAMESPACES=STATIC_CONFIG,ROLES

 #==================================================#
 #                      Others                      #
@@ -532,9 +654,9 @@ HELP_AND_FAQ_URL=https://librechat.ai
 #   users always get the latest version. Customize    #
 #   only if you understand caching implications.      #

-# INDEX_HTML_CACHE_CONTROL=no-cache, no-store, must-revalidate
-# INDEX_HTML_PRAGMA=no-cache
-# INDEX_HTML_EXPIRES=0
+# INDEX_CACHE_CONTROL=no-cache, no-store, must-revalidate
+# INDEX_PRAGMA=no-cache
+# INDEX_EXPIRES=0

 # no-cache: Forces validation with server before using cached version
 # no-store: Prevents storing the response entirely
@@ -544,3 +666,33 @@ HELP_AND_FAQ_URL=https://librechat.ai
 #                  OpenWeather                        #
 #=====================================================#
 OPENWEATHER_API_KEY=
+
+#====================================#
+# LibreChat Code Interpreter API     #
+#====================================#
+
+# https://code.librechat.ai
+# LIBRECHAT_CODE_API_KEY=your-key
+
+#======================#
+# Web Search           #
+#======================#
+
+# Note: All of the following variable names can be customized.
+# Omit values to allow user to provide them.
+
+# For more information on configuration values, see:
+# https://librechat.ai/docs/features/web_search
+
+# Search Provider (Required)
+# SERPER_API_KEY=your_serper_api_key
+
+# Scraper (Required)
+# FIRECRAWL_API_KEY=your_firecrawl_api_key
+# Optional: Custom Firecrawl API URL
+# FIRECRAWL_API_URL=your_firecrawl_api_url
+
+# Reranker (Required)
+# JINA_API_KEY=your_jina_api_key
+# or
+# COHERE_API_KEY=your_cohere_api_key
--- a/.github/CONTRIBUTING.md
+++ b/.github/CONTRIBUTING.md
@@ -24,22 +24,40 @@ Project maintainers have the right and responsibility to remove, edit, or reject

 ## To contribute to this project, please adhere to the following guidelines:

-## 1. Development notes
+## 1. Development Setup

-1. Before starting work, make sure your main branch has the latest commits with `npm run update`
-2. Run linting command to find errors: `npm run lint`. Alternatively, ensure husky pre-commit checks are functioning.
+1. Use Node.JS 20.x.
+2. Install typescript globally: `npm i -g typescript`.
+3. Run `npm ci` to install dependencies.
+4. Build the data provider: `npm run build:data-provider`.
+5. Build data schemas: `npm run build:data-schemas`.
+6. Build API methods: `npm run build:api`.
+7. Setup and run unit tests:
+    - Copy `.env.test`: `cp api/test/.env.test.example api/test/.env.test`.
+    - Run backend unit tests: `npm run test:api`.
+    - Run frontend unit tests: `npm run test:client`.
+8. Setup and run integration tests:
+    - Build client: `cd client && npm run build`.
+    - Create `.env`: `cp .env.example .env`.
+    - Install [MongoDB Community Edition](https://www.mongodb.com/docs/manual/administration/install-community/), ensure that `mongosh` connects to your local instance.
+    - Run: `npx install playwright`, then `npx playwright install`.
+    - Copy `config.local`: `cp e2e/config.local.example.ts e2e/config.local.ts`.
+    - Copy `librechat.yaml`: `cp librechat.example.yaml librechat.yaml`.
+    - Run: `npm run e2e`.
+
+## 2. Development Notes
+
+1. Before starting work, make sure your main branch has the latest commits with `npm run update`.
+3. Run linting command to find errors: `npm run lint`. Alternatively, ensure husky pre-commit checks are functioning.
 3. After your changes, reinstall packages in your current branch using `npm run reinstall` and ensure everything still works. 
    - Restart the ESLint server ("ESLint: Restart ESLint Server" in VS Code command bar) and your IDE after reinstalling or updating.
 4. Clear web app localStorage and cookies before and after changes.
-5. For frontend changes:
-    - Install typescript globally: `npm i -g typescript`.
-    - Compile typescript before and after changes to check for introduced errors: `cd client && tsc --noEmit`.
-6. Run tests locally:
-    - Backend unit tests: `npm run test:api`
-    - Frontend unit tests: `npm run test:client`
-    - Integration tests: `npm run e2e` (requires playwright installed, `npx install playwright`)
+5. For frontend changes, compile typescript before and after changes to check for introduced errors: `cd client && npm run build`.
+6. Run backend unit tests: `npm run test:api`.
+7. Run frontend unit tests: `npm run test:client`.
+8. Run integration tests: `npm run e2e`.

-## 2. Git Workflow
+## 3. Git Workflow

 We utilize a GitFlow workflow to manage changes to this project's codebase. Follow these general steps when contributing code:

@@ -49,7 +67,7 @@ We utilize a GitFlow workflow to manage changes to this project's codebase. Foll
 4. Submit a pull request with a clear and concise description of your changes and the reasons behind them.
 5. We will review your pull request, provide feedback as needed, and eventually merge the approved changes into the main branch.

-## 3. Commit Message Format
+## 4. Commit Message Format

 We follow the [semantic format](https://gist.github.com/joshbuchea/6f47e86d2510bce28f8e7f42ae84c716) for commit messages.

@@ -76,7 +94,7 @@ feat: add hat wobble
 ```


-## 4. Pull Request Process
+## 5. Pull Request Process

 When submitting a pull request, please follow these guidelines:

@@ -91,7 +109,7 @@ Ensure that your changes meet the following criteria:
 - The commit history is clean and easy to follow. You can use `git rebase` or `git merge --squash` to clean your commit history before submitting the pull request.
 - The pull request description clearly outlines the changes and the reasons behind them. Be sure to include the steps to test the pull request.

-## 5. Naming Conventions
+## 6. Naming Conventions

 Apply the following naming conventions to branches, labels, and other Git-related entities:

@@ -100,7 +118,7 @@ Apply the following naming conventions to branches, labels, and other Git-relate
 - **JS/TS:** Directories and file names: Descriptive and camelCase. First letter uppercased for React files (e.g., `helperFunction.ts, ReactComponent.tsx`).
 - **Docs:** Directories and file names: Descriptive and snake_case (e.g., `config_files.md`).

-## 6. TypeScript Conversion
+## 7. TypeScript Conversion

 1. **Original State**: The project was initially developed entirely in JavaScript (JS).

@@ -126,7 +144,7 @@ Apply the following naming conventions to branches, labels, and other Git-relate
      
   - **Current Stance**: At present, this backend transition is of lower priority and might not be pursued.

-## 7. Module Import Conventions
+## 8. Module Import Conventions

 - `npm` packages first, 
     - from shortest line (top) to longest (bottom)
--- a/.github/ISSUE_TEMPLATE/BUG-REPORT.yml
+++ b/.github/ISSUE_TEMPLATE/BUG-REPORT.yml
@@ -79,6 +79,8 @@ body:

        For UI-related issues, browser console logs can be very helpful. You can provide these as screenshots or paste the text here.
      render: shell
+    validations:
+      required: true
  - type: textarea
    id: screenshots
    attributes:
--- a/.github/workflows/backend-review.yml
+++ b/.github/workflows/backend-review.yml
@@ -7,6 +7,7 @@ on:
      - release/*
    paths:
      - 'api/**'
+      - 'packages/**'
 jobs:
  tests_Backend:
    name: Run Backend unit tests
@@ -36,8 +37,11 @@ jobs:
      - name: Install Data Provider Package
        run: npm run build:data-provider

-      - name: Install MCP Package
-        run: npm run build:mcp
+      - name: Install Data Schemas Package
+        run: npm run build:data-schemas
+
+      - name: Install API Package
+        run: npm run build:api

      - name: Create empty auth.json file
        run: |
@@ -61,4 +65,10 @@ jobs:
        run: cd api && npm run test:ci

      - name: Run librechat-data-provider unit tests
-        run: cd packages/data-provider && npm run test:ci
+        run: cd packages/data-provider && npm run test:ci
+
+      - name: Run @librechat/data-schemas unit tests
+        run: cd packages/data-schemas && npm run test:ci
+
+      - name: Run @librechat/api unit tests
+        run: cd packages/api && npm run test:ci
--- a/.github/workflows/client.yml
+++ b/.github/workflows/client.yml
@@ -0,0 +1,32 @@
+name: Publish `@librechat/client` to NPM
+
+on:
+  workflow_dispatch:
+    inputs:
+      reason:
+        description: 'Reason for manual trigger'
+        required: false
+        default: 'Manual publish requested'
+
+jobs:
+  build-and-publish:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      
+      - name: Use Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: '18.x'
+          
+      - name: Check if client package exists
+        run: |
+          if [ -d "packages/client" ]; then
+            echo "Client package directory found"
+          else
+            echo "Client package directory not found - workflow ready for future use"
+            exit 0
+          fi
+          
+      - name: Placeholder for future publishing
+        run: echo "Client package publishing workflow is ready" 
--- a/.github/workflows/data-schemas.yml
+++ b/.github/workflows/data-schemas.yml
@@ -0,0 +1,58 @@
+name: Publish `@librechat/data-schemas` to NPM
+
+on:
+  push:
+    branches:
+      - main
+    paths:
+      - 'packages/data-schemas/package.json'
+  workflow_dispatch:
+    inputs:
+      reason:
+        description: 'Reason for manual trigger'
+        required: false
+        default: 'Manual publish requested'
+
+jobs:
+  build-and-publish:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      
+      - name: Use Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: '18.x'
+          
+      - name: Install dependencies
+        run: cd packages/data-schemas && npm ci
+        
+      - name: Build
+        run: cd packages/data-schemas && npm run build
+        
+      - name: Set up npm authentication
+        run: echo "//registry.npmjs.org/:_authToken=${{ secrets.PUBLISH_NPM_TOKEN }}" > ~/.npmrc
+        
+      - name: Check version change
+        id: check
+        working-directory: packages/data-schemas
+        run: |
+          PACKAGE_VERSION=$(node -p "require('./package.json').version")
+          PUBLISHED_VERSION=$(npm view @librechat/data-schemas version 2>/dev/null || echo "0.0.0")
+          if [ "$PACKAGE_VERSION" = "$PUBLISHED_VERSION" ]; then
+            echo "No version change, skipping publish"
+            echo "skip=true" >> $GITHUB_OUTPUT
+          else
+            echo "Version changed, proceeding with publish"
+            echo "skip=false" >> $GITHUB_OUTPUT
+          fi
+            
+      - name: Pack package
+        if: steps.check.outputs.skip != 'true'
+        working-directory: packages/data-schemas
+        run: npm pack
+        
+      - name: Publish
+        if: steps.check.outputs.skip != 'true'
+        working-directory: packages/data-schemas
+        run: npm publish *.tgz --access public
--- a/.github/workflows/deploy-dev.yml
+++ b/.github/workflows/deploy-dev.yml
@@ -2,7 +2,7 @@ name: Update Test Server

 on:
  workflow_run:
-    workflows: ["Docker Dev Images Build"]
+    workflows: ["Docker Dev Branch Images Build"]
    types:
      - completed
  workflow_dispatch:
@@ -12,7 +12,8 @@ jobs:
    runs-on: ubuntu-latest
    if: |
      github.repository == 'danny-avila/LibreChat' &&
-      (github.event_name == 'workflow_dispatch' || github.event.workflow_run.conclusion == 'success')
+      (github.event_name == 'workflow_dispatch' || 
+       (github.event.workflow_run.conclusion == 'success' && github.event.workflow_run.head_branch == 'dev'))
    steps:
    - name: Checkout repository
      uses: actions/checkout@v4
@@ -29,13 +30,17 @@ jobs:
        DO_USER: ${{ secrets.DO_USER }}
      run: |
        ssh -o StrictHostKeyChecking=no ${DO_USER}@${DO_HOST} << EOF
-        sudo -i -u danny bash << EEOF
+        sudo -i -u danny bash << 'EEOF'
        cd ~/LibreChat && \
        git fetch origin main && \
-        npm run update:deployed && \
+        sudo npm run stop:deployed && \
+        sudo docker images --format "{{.Repository}}:{{.ID}}" | grep -E "lc-dev|librechat" | cut -d: -f2 | xargs -r sudo docker rmi -f || true && \
+        sudo npm run update:deployed && \
+        git checkout dev && \
+        git pull origin dev && \
        git checkout do-deploy && \
-        git rebase main && \
-        npm run start:deployed && \
+        git rebase dev && \
+        sudo npm run start:deployed && \
        echo "Update completed. Application should be running now."
        EEOF
        EOF
--- a/.github/workflows/dev-branch-images.yml
+++ b/.github/workflows/dev-branch-images.yml
@@ -0,0 +1,72 @@
+name: Docker Dev Branch Images Build
+
+on:
+  workflow_dispatch:
+  push:
+    branches:
+      - dev
+    paths:
+      - 'api/**'
+      - 'client/**'
+      - 'packages/**'
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        include:
+          - target: api-build
+            file: Dockerfile.multi
+            image_name: lc-dev-api
+          - target: node
+            file: Dockerfile
+            image_name: lc-dev
+
+    steps:
+      # Check out the repository
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      # Set up QEMU
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@v3
+
+      # Set up Docker Buildx
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
+
+      # Log in to GitHub Container Registry
+      - name: Log in to GitHub Container Registry
+        uses: docker/login-action@v2
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      # Login to Docker Hub
+      - name: Login to Docker Hub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ secrets.DOCKERHUB_USERNAME }}
+          password: ${{ secrets.DOCKERHUB_TOKEN }}
+
+      # Prepare the environment
+      - name: Prepare environment
+        run: |
+          cp .env.example .env
+
+      # Build and push Docker images for each target
+      - name: Build and push Docker images
+        uses: docker/build-push-action@v5
+        with:
+          context: .
+          file: ${{ matrix.file }}
+          push: true
+          tags: |
+            ghcr.io/${{ github.repository_owner }}/${{ matrix.image_name }}:${{ github.sha }}
+            ghcr.io/${{ github.repository_owner }}/${{ matrix.image_name }}:latest
+            ${{ secrets.DOCKERHUB_USERNAME }}/${{ matrix.image_name }}:${{ github.sha }}
+            ${{ secrets.DOCKERHUB_USERNAME }}/${{ matrix.image_name }}:latest
+          platforms: linux/amd64,linux/arm64
+          target: ${{ matrix.target }} 
--- a/.github/workflows/frontend-review.yml
+++ b/.github/workflows/frontend-review.yml
@@ -8,7 +8,7 @@ on:
      - release/*
    paths:
      - 'client/**'
-      - 'packages/**'
+      - 'packages/data-provider/**'

 jobs:
  tests_frontend_ubuntu:
--- a/.github/workflows/generate-release-changelog-pr.yml
+++ b/.github/workflows/generate-release-changelog-pr.yml
@@ -1,94 +0,0 @@
-name: Generate Release Changelog PR
-
-on:
-  push:
-    tags:
-      - 'v*.*.*'
-
-jobs:
-  generate-release-changelog-pr:
-    permissions:
-      contents: write    # Needed for pushing commits and creating branches.
-      pull-requests: write
-    runs-on: ubuntu-latest
-    steps:
-      # 1. Checkout the repository (with full history).
-      - name: Checkout Repository
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
-
-      # 2. Generate the release changelog using our custom configuration.
-      - name: Generate Release Changelog
-        id: generate_release
-        uses: mikepenz/release-changelog-builder-action@v5.1.0
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        with:
-          configuration: ".github/configuration-release.json"
-          owner: ${{ github.repository_owner }}
-          repo: ${{ github.event.repository.name }}
-          outputFile: CHANGELOG-release.md
-
-      # 3. Update the main CHANGELOG.md:
-      #    - If it doesn't exist, create it with a basic header.
-      #    - Remove the "Unreleased" section (if present).
-      #    - Prepend the new release changelog above previous releases.
-      #    - Remove all temporary files before committing.
-      - name: Update CHANGELOG.md
-        run: |
-          # Determine the release tag, e.g. "v1.2.3"
-          TAG=${GITHUB_REF##*/}
-          echo "Using release tag: $TAG"
-          
-          # Ensure CHANGELOG.md exists; if not, create a basic header.
-          if [ ! -f CHANGELOG.md ]; then
-            echo "# Changelog" > CHANGELOG.md
-            echo "" >> CHANGELOG.md
-            echo "All notable changes to this project will be documented in this file." >> CHANGELOG.md
-            echo "" >> CHANGELOG.md
-          fi
-
-          echo "Updating CHANGELOG.md…"
-          
-          # Remove the "Unreleased" section (from "## [Unreleased]" until the first occurrence of '---') if it exists.
-          if grep -q "^## \[Unreleased\]" CHANGELOG.md; then
-            awk '/^## \[Unreleased\]/{flag=1} flag && /^---/{flag=0; next} !flag' CHANGELOG.md > CHANGELOG.cleaned
-          else
-            cp CHANGELOG.md CHANGELOG.cleaned
-          fi
-
-          # Split the cleaned file into:
-          #   - header.md: content before the first release header ("## [v...").
-          #   - tail.md: content from the first release header onward.
-          awk '/^## \[v/{exit} {print}' CHANGELOG.cleaned > header.md
-          awk 'f{print} /^## \[v/{f=1; print}' CHANGELOG.cleaned > tail.md
-
-          # Combine header, the new release changelog, and the tail.
-          echo "Combining updated changelog parts..."
-          cat header.md CHANGELOG-release.md > CHANGELOG.md.new
-          echo "" >> CHANGELOG.md.new
-          cat tail.md >> CHANGELOG.md.new
-
-          mv CHANGELOG.md.new CHANGELOG.md
-
-          # Remove temporary files.
-          rm -f CHANGELOG.cleaned header.md tail.md CHANGELOG-release.md
-
-          echo "Final CHANGELOG.md content:"
-          cat CHANGELOG.md
-
-      # 4. Create (or update) the Pull Request with the updated CHANGELOG.md.
-      - name: Create Pull Request
-        uses: peter-evans/create-pull-request@v7
-        with:
-          token: ${{ secrets.GITHUB_TOKEN }}
-          sign-commits: true
-          commit-message: "chore: update CHANGELOG for release ${GITHUB_REF##*/}"
-          base: main
-          branch: "changelog/${GITHUB_REF##*/}"
-          reviewers: danny-avila
-          title: "chore: update CHANGELOG for release ${GITHUB_REF##*/}"
-          body: |
-            **Description**:
-            - This PR updates the CHANGELOG.md by removing the "Unreleased" section and adding new release notes for release ${GITHUB_REF##*/} above previous releases.
--- a/.github/workflows/generate-unreleased-changelog-pr.yml
+++ b/.github/workflows/generate-unreleased-changelog-pr.yml
@@ -1,106 +0,0 @@
-name: Generate Unreleased Changelog PR
-
-on:
-  schedule:
-    - cron: "0 0 * * 1"  # Runs every Monday at 00:00 UTC
-
-jobs:
-  generate-unreleased-changelog-pr:
-    permissions:
-      contents: write    # Needed for pushing commits and creating branches.
-      pull-requests: write
-    runs-on: ubuntu-latest
-    steps:
-      # 1. Checkout the repository on main.
-      - name: Checkout Repository on Main
-        uses: actions/checkout@v4
-        with:
-          ref: main
-          fetch-depth: 0
-
-      # 4. Get the latest version tag.
-      - name: Get Latest Tag
-        id: get_latest_tag
-        run: |
-          LATEST_TAG=$(git describe --tags $(git rev-list --tags --max-count=1) || echo "none")
-          echo "Latest tag: $LATEST_TAG"
-          echo "tag=$LATEST_TAG" >> $GITHUB_OUTPUT
-
-      # 5. Generate the Unreleased changelog.
-      - name: Generate Unreleased Changelog
-        id: generate_unreleased
-        uses: mikepenz/release-changelog-builder-action@v5.1.0
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        with:
-          configuration: ".github/configuration-unreleased.json"
-          owner: ${{ github.repository_owner }}
-          repo: ${{ github.event.repository.name }}
-          outputFile: CHANGELOG-unreleased.md
-          fromTag: ${{ steps.get_latest_tag.outputs.tag }}
-          toTag: main
-
-      # 7. Update CHANGELOG.md with the new Unreleased section.
-      - name: Update CHANGELOG.md
-        id: update_changelog
-        run: |
-          # Create CHANGELOG.md if it doesn't exist.
-          if [ ! -f CHANGELOG.md ]; then
-            echo "# Changelog" > CHANGELOG.md
-            echo "" >> CHANGELOG.md
-            echo "All notable changes to this project will be documented in this file." >> CHANGELOG.md
-            echo "" >> CHANGELOG.md
-          fi
-          
-          echo "Updating CHANGELOG.md…"
-          
-          # Extract content before the "## [Unreleased]" (or first version header if missing).
-          if grep -q "^## \[Unreleased\]" CHANGELOG.md; then
-            awk '/^## \[Unreleased\]/{exit} {print}' CHANGELOG.md > CHANGELOG_TMP.md
-          else
-            awk '/^## \[v/{exit} {print}' CHANGELOG.md > CHANGELOG_TMP.md
-          fi
-          
-          # Append the generated Unreleased changelog.
-          echo "" >> CHANGELOG_TMP.md
-          cat CHANGELOG-unreleased.md >> CHANGELOG_TMP.md
-          echo "" >> CHANGELOG_TMP.md
-          
-          # Append the remainder of the original changelog (starting from the first version header).
-          awk 'f{print} /^## \[v/{f=1; print}' CHANGELOG.md >> CHANGELOG_TMP.md
-          
-          # Replace the old file with the updated file.
-          mv CHANGELOG_TMP.md CHANGELOG.md
-          
-          # Remove the temporary generated file.
-          rm -f CHANGELOG-unreleased.md
-          
-          echo "Final CHANGELOG.md:"
-          cat CHANGELOG.md
-
-      # 8. Check if CHANGELOG.md has any updates.
-      - name: Check for CHANGELOG.md changes
-        id: changelog_changes
-        run: |
-          if git diff --quiet CHANGELOG.md; then
-            echo "has_changes=false" >> $GITHUB_OUTPUT
-          else
-            echo "has_changes=true" >> $GITHUB_OUTPUT
-          fi
-
-      # 9. Create (or update) the Pull Request only if there are changes.
-      - name: Create Pull Request
-        if: steps.changelog_changes.outputs.has_changes == 'true'
-        uses: peter-evans/create-pull-request@v7
-        with:
-          token: ${{ secrets.GITHUB_TOKEN }}
-          base: main
-          branch: "changelog/unreleased-update"
-          sign-commits: true
-          commit-message: "action: update Unreleased changelog"
-          title: "action: update Unreleased changelog"
-          body: |
-            **Description**:
-            - This PR updates the Unreleased section in CHANGELOG.md.
-            - It compares the current main branch with the latest version tag (determined as ${{ steps.get_latest_tag.outputs.tag }}),
-              regenerates the Unreleased changelog, removes any old Unreleased block, and inserts the new content.
--- a/.github/workflows/helmcharts.yml
+++ b/.github/workflows/helmcharts.yml
@@ -26,8 +26,15 @@ jobs:
        uses: azure/setup-helm@v4
        env:
          GITHUB_TOKEN: "${{ secrets.GITHUB_TOKEN }}"
+      - name: Build Subchart Deps
+        run: |
+          cd helm/librechat-rag-api
+          helm dependency build 

      - name: Run chart-releaser
        uses: helm/chart-releaser-action@v1.6.0
+        with:
+          charts_dir: helm
+          skip_existing: true
        env:
          CR_TOKEN: "${{ secrets.GITHUB_TOKEN }}"
--- a/.github/workflows/i18n-unused-keys.yml
+++ b/.github/workflows/i18n-unused-keys.yml
@@ -5,12 +5,13 @@ on:
    paths:
      - "client/src/**"
      - "api/**"
+      - "packages/data-provider/src/**"

 jobs:
  detect-unused-i18n-keys:
    runs-on: ubuntu-latest
    permissions:
-      pull-requests: write  # Required for posting PR comments
+      pull-requests: write
    steps:
      - name: Checkout repository
        uses: actions/checkout@v3
@@ -22,7 +23,7 @@ jobs:

          # Define paths
          I18N_FILE="client/src/locales/en/translation.json"
-          SOURCE_DIRS=("client/src" "api")
+          SOURCE_DIRS=("client/src" "api" "packages/data-provider/src")

          # Check if translation file exists
          if [[ ! -f "$I18N_FILE" ]]; then
@@ -39,12 +40,35 @@ jobs:
          # Check if each key is used in the source code
          for KEY in $KEYS; do
            FOUND=false
-            for DIR in "${SOURCE_DIRS[@]}"; do
-              if grep -r --include=\*.{js,jsx,ts,tsx} -q "$KEY" "$DIR"; then
-                FOUND=true
-                break
+            
+            # Special case for dynamically constructed special variable keys
+            if [[ "$KEY" == com_ui_special_var_* ]]; then
+              # Check if TSpecialVarLabel is used in the codebase
+              for DIR in "${SOURCE_DIRS[@]}"; do
+                if grep -r --include=\*.{js,jsx,ts,tsx} -q "TSpecialVarLabel" "$DIR"; then
+                  FOUND=true
+                  break
+                fi
+              done
+              
+              # Also check if the key is directly used somewhere
+              if [[ "$FOUND" == false ]]; then
+                for DIR in "${SOURCE_DIRS[@]}"; do
+                  if grep -r --include=\*.{js,jsx,ts,tsx} -q "$KEY" "$DIR"; then
+                    FOUND=true
+                    break
+                  fi
+                done
              fi
-            done
+            else
+              # Regular check for other keys
+              for DIR in "${SOURCE_DIRS[@]}"; do
+                if grep -r --include=\*.{js,jsx,ts,tsx} -q "$KEY" "$DIR"; then
+                  FOUND=true
+                  break
+                fi
+              done
+            fi
          
            if [[ "$FOUND" == false ]]; then
              UNUSED_KEYS+=("$KEY")
@@ -90,4 +114,4 @@ jobs:

      - name: Fail workflow if unused keys found
        if: env.unused_keys != '[]'
-        run: exit 1
+        run: exit 1
--- a/.github/workflows/unused-packages.yml
+++ b/.github/workflows/unused-packages.yml
@@ -1,6 +1,12 @@
 name: Detect Unused NPM Packages

-on: [pull_request]
+on:
+  pull_request:
+    paths:
+      - 'package.json'
+      - 'package-lock.json'
+      - 'client/**'
+      - 'api/**'

 jobs:
  detect-unused-packages:
@@ -73,12 +79,52 @@ jobs:
          extract_deps_from_code "client" client_used_code.txt
          extract_deps_from_code "api" api_used_code.txt

+      - name: Extract Workspace Dependencies
+        id: extract-workspace-deps
+        run: |
+          # Function to get dependencies from a workspace package that are used by another package
+          get_workspace_package_deps() {
+            local package_json=$1
+            local output_file=$2
+            
+            # Get all workspace dependencies (starting with @librechat/)
+            if [[ -f "$package_json" ]]; then
+              local workspace_deps=$(jq -r '.dependencies // {} | to_entries[] | select(.key | startswith("@librechat/")) | .key' "$package_json" 2>/dev/null || echo "")
+              
+              # For each workspace dependency, get its dependencies
+              for dep in $workspace_deps; do
+                # Convert @librechat/api to packages/api
+                local workspace_path=$(echo "$dep" | sed 's/@librechat\//packages\//')
+                local workspace_package_json="${workspace_path}/package.json"
+                
+                if [[ -f "$workspace_package_json" ]]; then
+                  # Extract all dependencies from the workspace package
+                  jq -r '.dependencies // {} | keys[]' "$workspace_package_json" 2>/dev/null >> "$output_file"
+                  # Also extract peerDependencies
+                  jq -r '.peerDependencies // {} | keys[]' "$workspace_package_json" 2>/dev/null >> "$output_file"
+                fi
+              done
+            fi
+            
+            if [[ -f "$output_file" ]]; then
+              sort -u "$output_file" -o "$output_file"
+            else
+              touch "$output_file"
+            fi
+          }
+          
+          # Get workspace dependencies for each package
+          get_workspace_package_deps "package.json" root_workspace_deps.txt
+          get_workspace_package_deps "client/package.json" client_workspace_deps.txt
+          get_workspace_package_deps "api/package.json" api_workspace_deps.txt
+
      - name: Run depcheck for root package.json
        id: check-root
        run: |
          if [[ -f "package.json" ]]; then
            UNUSED=$(depcheck --json | jq -r '.dependencies | join("\n")' || echo "")
-            UNUSED=$(comm -23 <(echo "$UNUSED" | sort) <(cat root_used_deps.txt root_used_code.txt | sort) || echo "")
+            # Exclude dependencies used in scripts, code, and workspace packages
+            UNUSED=$(comm -23 <(echo "$UNUSED" | sort) <(cat root_used_deps.txt root_used_code.txt root_workspace_deps.txt | sort) || echo "")
            echo "ROOT_UNUSED<<EOF" >> $GITHUB_ENV
            echo "$UNUSED" >> $GITHUB_ENV
            echo "EOF" >> $GITHUB_ENV
@@ -91,7 +137,10 @@ jobs:
            chmod -R 755 client
            cd client
            UNUSED=$(depcheck --json | jq -r '.dependencies | join("\n")' || echo "")
-            UNUSED=$(comm -23 <(echo "$UNUSED" | sort) <(cat ../client_used_deps.txt ../client_used_code.txt | sort) || echo "")
+            # Exclude dependencies used in scripts, code, and workspace packages
+            UNUSED=$(comm -23 <(echo "$UNUSED" | sort) <(cat ../client_used_deps.txt ../client_used_code.txt ../client_workspace_deps.txt | sort) || echo "")
+            # Filter out false positives
+            UNUSED=$(echo "$UNUSED" | grep -v "^micromark-extension-llm-math$" || echo "")
            echo "CLIENT_UNUSED<<EOF" >> $GITHUB_ENV
            echo "$UNUSED" >> $GITHUB_ENV
            echo "EOF" >> $GITHUB_ENV
@@ -105,7 +154,8 @@ jobs:
            chmod -R 755 api
            cd api
            UNUSED=$(depcheck --json | jq -r '.dependencies | join("\n")' || echo "")
-            UNUSED=$(comm -23 <(echo "$UNUSED" | sort) <(cat ../api_used_deps.txt ../api_used_code.txt | sort) || echo "")
+            # Exclude dependencies used in scripts, code, and workspace packages
+            UNUSED=$(comm -23 <(echo "$UNUSED" | sort) <(cat ../api_used_deps.txt ../api_used_code.txt ../api_workspace_deps.txt | sort) || echo "")
            echo "API_UNUSED<<EOF" >> $GITHUB_ENV
            echo "$UNUSED" >> $GITHUB_ENV
            echo "EOF" >> $GITHUB_ENV
--- a/.gitignore
+++ b/.gitignore
@@ -37,6 +37,10 @@ client/public/main.js
 client/public/main.js.map
 client/public/main.js.LICENSE.txt

+# Azure Blob Storage Emulator (Azurite)
+__azurite**
+__blobstorage__/**/*
+
 # Dependency directorys
 # Deployed apps should consider commenting these lines out:
 # see https://npmjs.org/doc/faq.html#Should-I-check-my-node_modules-folder-into-git
@@ -48,6 +52,11 @@ bower_components/
 *.d.ts
 !vite-env.d.ts

+# AI
+.clineignore
+.cursor
+.aider*
+
 # Floobits
 .floo
 .floobit
@@ -106,4 +115,22 @@ uploads/

 # owner
 release/
+
+# Helm
+helm/librechat/Chart.lock
+helm/**/charts/
+helm/**/.values.yaml
+
 !/client/src/@types/i18next.d.ts
+
+# SAML Idp cert
+*.cert
+
+# AI Assistants
+/.claude/
+/.cursor/
+/.copilot/
+/.aider/
+/.openai/
+/.tabnine/
+/.codeium
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -0,0 +1,236 @@
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+
+
+
+
+
+## [Unreleased]
+
+### ✨ New Features
+
+- ✨ feat: implement search parameter updates by **@mawburn** in [#7151](https://github.com/danny-avila/LibreChat/pull/7151)
+- 🎏 feat: Add MCP support for Streamable HTTP Transport by **@benverhees** in [#7353](https://github.com/danny-avila/LibreChat/pull/7353)
+- 🔒 feat: Add Content Security Policy using Helmet middleware by **@rubentalstra** in [#7377](https://github.com/danny-avila/LibreChat/pull/7377)
+- ✨ feat: Add Normalization for MCP Server Names by **@danny-avila** in [#7421](https://github.com/danny-avila/LibreChat/pull/7421)
+- 📊 feat: Improve Helm Chart by **@hofq** in [#3638](https://github.com/danny-avila/LibreChat/pull/3638)
+- 🦾 feat: Claude-4 Support by **@danny-avila** in [#7509](https://github.com/danny-avila/LibreChat/pull/7509)
+- 🪨 feat: Bedrock Support for Claude-4 Reasoning by **@danny-avila** in [#7517](https://github.com/danny-avila/LibreChat/pull/7517)
+
+### 🌍 Internationalization
+
+- 🌍 i18n: Add `Danish` and `Czech` and `Catalan` localization support by **@rubentalstra** in [#7373](https://github.com/danny-avila/LibreChat/pull/7373)
+- 🌍 i18n: Update translation.json with latest translations by **@github-actions[bot]** in [#7375](https://github.com/danny-avila/LibreChat/pull/7375)
+- 🌍 i18n: Update translation.json with latest translations by **@github-actions[bot]** in [#7468](https://github.com/danny-avila/LibreChat/pull/7468)
+
+### 🔧 Fixes
+
+- 💬 fix: update aria-label for accessibility in ConvoLink component by **@berry-13** in [#7320](https://github.com/danny-avila/LibreChat/pull/7320)
+- 🔑 fix: use `apiKey` instead of `openAIApiKey` in OpenAI-like Config by **@danny-avila** in [#7337](https://github.com/danny-avila/LibreChat/pull/7337)
+- 🔄 fix: update navigation logic in `useFocusChatEffect` to ensure correct search parameters are used by **@mawburn** in [#7340](https://github.com/danny-avila/LibreChat/pull/7340)
+- 🔄 fix: Improve MCP Connection Cleanup by **@danny-avila** in [#7400](https://github.com/danny-avila/LibreChat/pull/7400)
+- 🛡️ fix: Preset and Validation Logic for URL Query Params by **@danny-avila** in [#7407](https://github.com/danny-avila/LibreChat/pull/7407)
+- 🌘 fix: artifact of preview text is illegible in dark mode by **@nhtruong** in [#7405](https://github.com/danny-avila/LibreChat/pull/7405)
+- 🛡️ fix: Temporarily Remove CSP until Configurable by **@danny-avila** in [#7419](https://github.com/danny-avila/LibreChat/pull/7419)
+- 💽 fix: Exclude index page `/` from static cache settings by **@sbruel** in [#7382](https://github.com/danny-avila/LibreChat/pull/7382)
+
+### ⚙️ Other Changes
+
+- 📜 docs: CHANGELOG for release v0.7.8 by **@github-actions[bot]** in [#7290](https://github.com/danny-avila/LibreChat/pull/7290)
+- 📦 chore: Update API Package Dependencies by **@danny-avila** in [#7359](https://github.com/danny-avila/LibreChat/pull/7359)
+- 📜 docs: Unreleased Changelog by **@github-actions[bot]** in [#7321](https://github.com/danny-avila/LibreChat/pull/7321)
+- 📜 docs: Unreleased Changelog by **@github-actions[bot]** in [#7434](https://github.com/danny-avila/LibreChat/pull/7434)
+- 🛡️ chore: `multer` v2.0.0 for CVE-2025-47935 and CVE-2025-47944 by **@danny-avila** in [#7454](https://github.com/danny-avila/LibreChat/pull/7454)
+- 📂 refactor: Improve `FileAttachment` & File Form Deletion by **@danny-avila** in [#7471](https://github.com/danny-avila/LibreChat/pull/7471)
+- 📊 chore: Remove Old Helm Chart by **@hofq** in [#7512](https://github.com/danny-avila/LibreChat/pull/7512)
+- 🪖 chore: bump helm app version to v0.7.8 by **@austin-barrington** in [#7524](https://github.com/danny-avila/LibreChat/pull/7524)
+
+
+
+---
+## [v0.7.8] - 
+
+Changes from v0.7.8-rc1 to v0.7.8.
+
+### ✨ New Features
+
+- ✨ feat: Enhance form submission for touch screens by **@berry-13** in [#7198](https://github.com/danny-avila/LibreChat/pull/7198)
+- 🔍 feat: Additional Tavily API Tool Parameters by **@glowforge-opensource** in [#7232](https://github.com/danny-avila/LibreChat/pull/7232)
+- 🐋 feat: Add python to Dockerfile for increased MCP compatibility by **@technicalpickles** in [#7270](https://github.com/danny-avila/LibreChat/pull/7270)
+
+### 🔧 Fixes
+
+- 🔧 fix: Google Gemma Support & OpenAI Reasoning Instructions by **@danny-avila** in [#7196](https://github.com/danny-avila/LibreChat/pull/7196)
+- 🛠️ fix: Conversation Navigation State by **@danny-avila** in [#7210](https://github.com/danny-avila/LibreChat/pull/7210)
+- 🔄 fix: o-Series Model Regex for System Messages by **@danny-avila** in [#7245](https://github.com/danny-avila/LibreChat/pull/7245)
+- 🔖 fix: Custom Headers for Initial MCP SSE Connection by **@danny-avila** in [#7246](https://github.com/danny-avila/LibreChat/pull/7246)
+- 🛡️ fix: Deep Clone `MCPOptions` for User MCP Connections by **@danny-avila** in [#7247](https://github.com/danny-avila/LibreChat/pull/7247)
+- 🔄 fix: URL Param Race Condition and File Draft Persistence by **@danny-avila** in [#7257](https://github.com/danny-avila/LibreChat/pull/7257)
+- 🔄 fix: Assistants Endpoint & Minor Issues by **@danny-avila** in [#7274](https://github.com/danny-avila/LibreChat/pull/7274)
+- 🔄 fix: Ollama Think Tag Edge Case with Tools by **@danny-avila** in [#7275](https://github.com/danny-avila/LibreChat/pull/7275)
+
+### ⚙️ Other Changes
+
+- 📜 docs: CHANGELOG for release v0.7.8-rc1 by **@github-actions[bot]** in [#7153](https://github.com/danny-avila/LibreChat/pull/7153)
+- 🔄 refactor: Artifact Visibility Management by **@danny-avila** in [#7181](https://github.com/danny-avila/LibreChat/pull/7181)
+- 📦 chore: Bump Package Security by **@danny-avila** in [#7183](https://github.com/danny-avila/LibreChat/pull/7183)
+- 🌿 refactor: Unmount Fork Popover on Hide for Better Performance by **@danny-avila** in [#7189](https://github.com/danny-avila/LibreChat/pull/7189)
+- 🧰 chore: ESLint configuration to enforce Prettier formatting rules by **@mawburn** in [#7186](https://github.com/danny-avila/LibreChat/pull/7186)
+- 🎨 style: Improve KaTeX Rendering for LaTeX Equations by **@andresgit** in [#7223](https://github.com/danny-avila/LibreChat/pull/7223)
+- 📝 docs: Update `.env.example` Google models by **@marlonka** in [#7254](https://github.com/danny-avila/LibreChat/pull/7254)
+- 💬 refactor: MCP Chat Visibility Option, Google Rates, Remove OpenAPI Plugins by **@danny-avila** in [#7286](https://github.com/danny-avila/LibreChat/pull/7286)
+- 📜 docs: Unreleased Changelog by **@github-actions[bot]** in [#7214](https://github.com/danny-avila/LibreChat/pull/7214)
+
+
+
+[See full release details][release-v0.7.8]
+
+[release-v0.7.8]: https://github.com/danny-avila/LibreChat/releases/tag/v0.7.8
+
+---
+## [v0.7.8-rc1] - 
+
+Changes from v0.7.7 to v0.7.8-rc1.
+
+### ✨ New Features
+
+- 🔍 feat: Mistral OCR API / Upload Files as Text by **@danny-avila** in [#6274](https://github.com/danny-avila/LibreChat/pull/6274)
+- 🤖 feat: Support OpenAI Web Search models by **@danny-avila** in [#6313](https://github.com/danny-avila/LibreChat/pull/6313)
+- 🔗 feat: Agent Chain (Mixture-of-Agents) by **@danny-avila** in [#6374](https://github.com/danny-avila/LibreChat/pull/6374)
+- ⌛ feat: `initTimeout` for Slow Starting MCP Servers by **@perweij** in [#6383](https://github.com/danny-avila/LibreChat/pull/6383)
+- 🚀 feat: `S3` Integration for File handling and Image uploads by **@rubentalstra** in [#6142](https://github.com/danny-avila/LibreChat/pull/6142)
+- 🔒feat: Enable OpenID Auto-Redirect by **@leondape** in [#6066](https://github.com/danny-avila/LibreChat/pull/6066)
+- 🚀 feat: Integrate `Azure Blob Storage` for file handling and image uploads by **@rubentalstra** in [#6153](https://github.com/danny-avila/LibreChat/pull/6153)
+- 🚀 feat: Add support for custom `AWS` endpoint in `S3` by **@rubentalstra** in [#6431](https://github.com/danny-avila/LibreChat/pull/6431)
+- 🚀 feat: Add support for LDAP STARTTLS in LDAP authentication by **@rubentalstra** in [#6438](https://github.com/danny-avila/LibreChat/pull/6438)
+- 🚀 feat: Refactor schema exports and update package version to 0.0.4 by **@rubentalstra** in [#6455](https://github.com/danny-avila/LibreChat/pull/6455)
+- 🔼 feat: Add Auto Submit For URL Query Params by **@mjaverto** in [#6440](https://github.com/danny-avila/LibreChat/pull/6440)
+- 🛠 feat: Enhance Redis Integration, Rate Limiters & Log Headers by **@danny-avila** in [#6462](https://github.com/danny-avila/LibreChat/pull/6462)
+- 💵 feat: Add Automatic Balance Refill by **@rubentalstra** in [#6452](https://github.com/danny-avila/LibreChat/pull/6452)
+- 🗣️ feat: add support for gpt-4o-transcribe models by **@berry-13** in [#6483](https://github.com/danny-avila/LibreChat/pull/6483)
+- 🎨 feat: UI Refresh for Enhanced UX by **@berry-13** in [#6346](https://github.com/danny-avila/LibreChat/pull/6346)
+- 🌍 feat: Add support for Hungarian language localization by **@rubentalstra** in [#6508](https://github.com/danny-avila/LibreChat/pull/6508)
+- 🚀 feat: Add Gemini 2.5 Token/Context Values, Increase Max Possible Output to 64k by **@danny-avila** in [#6563](https://github.com/danny-avila/LibreChat/pull/6563)
+- 🚀 feat: Enhance MCP Connections For Multi-User Support by **@danny-avila** in [#6610](https://github.com/danny-avila/LibreChat/pull/6610)
+- 🚀 feat: Enhance S3 URL Expiry with Refresh; fix: S3 File Deletion by **@danny-avila** in [#6647](https://github.com/danny-avila/LibreChat/pull/6647)
+- 🚀 feat: enhance UI components and refactor settings by **@berry-13** in [#6625](https://github.com/danny-avila/LibreChat/pull/6625)
+- 💬 feat: move TemporaryChat to the Header by **@berry-13** in [#6646](https://github.com/danny-avila/LibreChat/pull/6646)
+- 🚀 feat: Use Model Specs + Specific Endpoints, Limit Providers for Agents by **@danny-avila** in [#6650](https://github.com/danny-avila/LibreChat/pull/6650)
+- 🪙 feat: Sync Balance Config on Login by **@danny-avila** in [#6671](https://github.com/danny-avila/LibreChat/pull/6671)
+- 🔦 feat: MCP Support for Non-Agent Endpoints by **@danny-avila** in [#6775](https://github.com/danny-avila/LibreChat/pull/6775)
+- 🗃️ feat: Code Interpreter File Persistence between Sessions by **@danny-avila** in [#6790](https://github.com/danny-avila/LibreChat/pull/6790)
+- 🖥️ feat: Code Interpreter API for Non-Agent Endpoints by **@danny-avila** in [#6803](https://github.com/danny-avila/LibreChat/pull/6803)
+- ⚡ feat: Self-hosted Artifacts Static Bundler URL by **@danny-avila** in [#6827](https://github.com/danny-avila/LibreChat/pull/6827)
+- 🐳 feat: Add Jemalloc and UV to Docker Builds by **@danny-avila** in [#6836](https://github.com/danny-avila/LibreChat/pull/6836)
+- 🤖 feat: GPT-4.1 by **@danny-avila** in [#6880](https://github.com/danny-avila/LibreChat/pull/6880)
+- 👋 feat: remove Edge TTS by **@berry-13** in [#6885](https://github.com/danny-avila/LibreChat/pull/6885)
+- feat: nav optimization  by **@berry-13** in [#5785](https://github.com/danny-avila/LibreChat/pull/5785)
+- 🗺️ feat: Add Parameter Location Mapping for OpenAPI actions by **@peeeteeer** in [#6858](https://github.com/danny-avila/LibreChat/pull/6858)
+- 🤖 feat: Support `o4-mini` and `o3` Models by **@danny-avila** in [#6928](https://github.com/danny-avila/LibreChat/pull/6928)
+- 🎨 feat: OpenAI Image Tools (GPT-Image-1) by **@danny-avila** in [#7079](https://github.com/danny-avila/LibreChat/pull/7079)
+- 🗓️ feat: Add Special Variables for Prompts & Agents, Prompt UI Improvements by **@danny-avila** in [#7123](https://github.com/danny-avila/LibreChat/pull/7123)
+
+### 🌍 Internationalization
+
+- 🌍 i18n: Add Thai Language Support and Update Translations by **@rubentalstra** in [#6219](https://github.com/danny-avila/LibreChat/pull/6219)
+- 🌍 i18n: Update translation.json with latest translations by **@github-actions[bot]** in [#6220](https://github.com/danny-avila/LibreChat/pull/6220)
+- 🌍 i18n: Update translation.json with latest translations by **@github-actions[bot]** in [#6240](https://github.com/danny-avila/LibreChat/pull/6240)
+- 🌍 i18n: Update translation.json with latest translations by **@github-actions[bot]** in [#6241](https://github.com/danny-avila/LibreChat/pull/6241)
+- 🌍 i18n: Update translation.json with latest translations by **@github-actions[bot]** in [#6277](https://github.com/danny-avila/LibreChat/pull/6277)
+- 🌍 i18n: Update translation.json with latest translations by **@github-actions[bot]** in [#6414](https://github.com/danny-avila/LibreChat/pull/6414)
+- 🌍 i18n: Update translation.json with latest translations by **@github-actions[bot]** in [#6505](https://github.com/danny-avila/LibreChat/pull/6505)
+- 🌍 i18n: Update translation.json with latest translations by **@github-actions[bot]** in [#6530](https://github.com/danny-avila/LibreChat/pull/6530)
+- 🌍 i18n: Add Persian Localization Support by **@rubentalstra** in [#6669](https://github.com/danny-avila/LibreChat/pull/6669)
+- 🌍 i18n: Update translation.json with latest translations by **@github-actions[bot]** in [#6667](https://github.com/danny-avila/LibreChat/pull/6667)
+- 🌍 i18n: Update translation.json with latest translations by **@github-actions[bot]** in [#7126](https://github.com/danny-avila/LibreChat/pull/7126)
+- 🌍 i18n: Update translation.json with latest translations by **@github-actions[bot]** in [#7148](https://github.com/danny-avila/LibreChat/pull/7148)
+
+### 👐 Accessibility
+
+- 🎨 a11y: Update Model Spec Description Text by **@berry-13** in [#6294](https://github.com/danny-avila/LibreChat/pull/6294)
+- 🗑️ a11y: Add Accessible Name to Button for File Attachment Removal by **@kangabell** in [#6709](https://github.com/danny-avila/LibreChat/pull/6709)
+- ⌨️ a11y: enhance accessibility & visual consistency by **@berry-13** in [#6866](https://github.com/danny-avila/LibreChat/pull/6866)
+- 🙌 a11y: Searchbar/Conversations List Focus by **@danny-avila** in [#7096](https://github.com/danny-avila/LibreChat/pull/7096)
+- 👐 a11y: Improve Fork and SplitText Accessibility by **@danny-avila** in [#7147](https://github.com/danny-avila/LibreChat/pull/7147)
+
+### 🔧 Fixes
+
+- 🐛 fix: Avatar Type Definitions in Agent/Assistant Schemas by **@danny-avila** in [#6235](https://github.com/danny-avila/LibreChat/pull/6235)
+- 🔧 fix: MeiliSearch Field Error and Patch Incorrect Import by #6210 by **@rubentalstra** in [#6245](https://github.com/danny-avila/LibreChat/pull/6245)
+- 🔏 fix: Enhance Two-Factor Authentication by **@rubentalstra** in [#6247](https://github.com/danny-avila/LibreChat/pull/6247)
+- 🐛 fix: Await saveMessage in abortMiddleware to ensure proper execution by **@sh4shii** in [#6248](https://github.com/danny-avila/LibreChat/pull/6248)
+- 🔧 fix: Axios Proxy Usage And Bump `mongoose` by **@danny-avila** in [#6298](https://github.com/danny-avila/LibreChat/pull/6298)
+- 🔧 fix: comment out MCP servers to resolve service run issues by **@KunalScriptz** in [#6316](https://github.com/danny-avila/LibreChat/pull/6316)
+- 🔧 fix: Update Token Calculations and Mapping, MCP `env` Initialization by **@danny-avila** in [#6406](https://github.com/danny-avila/LibreChat/pull/6406)
+- 🐞 fix: Agent "Resend" Message Attachments + Source Icon Styling by **@danny-avila** in [#6408](https://github.com/danny-avila/LibreChat/pull/6408)
+- 🐛 fix: Prevent Crash on Duplicate Message ID by **@Odrec** in [#6392](https://github.com/danny-avila/LibreChat/pull/6392)
+- 🔐 fix: Invalid Key Length in 2FA Encryption by **@rubentalstra** in [#6432](https://github.com/danny-avila/LibreChat/pull/6432)
+- 🏗️ fix: Fix Agents Token Spend Race Conditions, Expand Test Coverage by **@danny-avila** in [#6480](https://github.com/danny-avila/LibreChat/pull/6480)
+- 🔃 fix: Draft Clearing, Claude Titles, Remove Default Vision Max Tokens by **@danny-avila** in [#6501](https://github.com/danny-avila/LibreChat/pull/6501)
+- 🔧 fix: Update username reference to use user.name in greeting display by **@rubentalstra** in [#6534](https://github.com/danny-avila/LibreChat/pull/6534)
+- 🔧 fix: S3 Download Stream with Key Extraction and Blob Storage Encoding for Vision by **@danny-avila** in [#6557](https://github.com/danny-avila/LibreChat/pull/6557)
+- 🔧 fix: Mistral type strictness for `usage` & update token values/windows by **@danny-avila** in [#6562](https://github.com/danny-avila/LibreChat/pull/6562)
+- 🔧 fix: Consolidate Text Parsing and TTS Edge Initialization by **@danny-avila** in [#6582](https://github.com/danny-avila/LibreChat/pull/6582)
+- 🔧 fix: Ensure continuation in image processing on base64 encoding from Blob Storage by **@danny-avila** in [#6619](https://github.com/danny-avila/LibreChat/pull/6619)
+- ✉️ fix: Fallback For User Name In Email Templates by **@danny-avila** in [#6620](https://github.com/danny-avila/LibreChat/pull/6620)
+- 🔧 fix: Azure Blob Integration and File Source References by **@rubentalstra** in [#6575](https://github.com/danny-avila/LibreChat/pull/6575)
+- 🐛 fix: Safeguard against undefined addedEndpoints by **@wipash** in [#6654](https://github.com/danny-avila/LibreChat/pull/6654)
+- 🤖 fix: Gemini 2.5 Vision Support by **@danny-avila** in [#6663](https://github.com/danny-avila/LibreChat/pull/6663)
+- 🔄 fix: Avatar & Error Handling Enhancements by **@danny-avila** in [#6687](https://github.com/danny-avila/LibreChat/pull/6687)
+- 🔧 fix: Chat Middleware, Zod Conversion, Auto-Save and S3 URL Refresh by **@danny-avila** in [#6720](https://github.com/danny-avila/LibreChat/pull/6720)
+- 🔧 fix: Agent Capability Checks & DocumentDB Compatibility for Agent Resource Removal by **@danny-avila** in [#6726](https://github.com/danny-avila/LibreChat/pull/6726)
+- 🔄 fix: Improve audio MIME type detection and handling by **@berry-13** in [#6707](https://github.com/danny-avila/LibreChat/pull/6707)
+- 🪺 fix: Update Role Handling due to New Schema Shape by **@danny-avila** in [#6774](https://github.com/danny-avila/LibreChat/pull/6774)
+- 🗨️ fix: Show ModelSpec Greeting by **@berry-13** in [#6770](https://github.com/danny-avila/LibreChat/pull/6770)
+- 🔧 fix: Keyv and Proxy Issues, and More Memory Optimizations by **@danny-avila** in [#6867](https://github.com/danny-avila/LibreChat/pull/6867)
+- ✨ fix: Implement dynamic text sizing for greeting and name display by **@berry-13** in [#6833](https://github.com/danny-avila/LibreChat/pull/6833)
+- 📝 fix: Mistral OCR Image Support and Azure Agent Titles by **@danny-avila** in [#6901](https://github.com/danny-avila/LibreChat/pull/6901)
+- 📢 fix: Invalid `engineTTS` and Conversation State on Navigation by **@berry-13** in [#6904](https://github.com/danny-avila/LibreChat/pull/6904)
+- 🛠️ fix: Improve Accessibility and Display of Conversation Menu by **@danny-avila** in [#6913](https://github.com/danny-avila/LibreChat/pull/6913)
+- 🔧 fix: Agent Resource Form, Convo Menu Style, Ensure Draft Clears on Submission by **@danny-avila** in [#6925](https://github.com/danny-avila/LibreChat/pull/6925)
+- 🔀 fix: MCP Improvements, Auto-Save Drafts, Artifact Markup by **@danny-avila** in [#7040](https://github.com/danny-avila/LibreChat/pull/7040)
+- 🐋 fix: Improve Deepseek Compatbility by **@danny-avila** in [#7132](https://github.com/danny-avila/LibreChat/pull/7132)
+- 🐙 fix: Add Redis Ping Interval to Prevent Connection Drops by **@peeeteeer** in [#7127](https://github.com/danny-avila/LibreChat/pull/7127)
+
+### ⚙️ Other Changes
+
+- 📦 refactor: Move DB Models to `@librechat/data-schemas` by **@rubentalstra** in [#6210](https://github.com/danny-avila/LibreChat/pull/6210)
+- 📦 chore: Patch `axios` to address CVE-2025-27152 by **@danny-avila** in [#6222](https://github.com/danny-avila/LibreChat/pull/6222)
+- ⚠️ refactor: Use Error Content Part Instead Of Throwing Error for Agents by **@danny-avila** in [#6262](https://github.com/danny-avila/LibreChat/pull/6262)
+- 🏃‍♂️ refactor: Improve Agent Run Context & Misc. Changes by **@danny-avila** in [#6448](https://github.com/danny-avila/LibreChat/pull/6448)
+- 📝 docs: librechat.example.yaml by **@ineiti** in [#6442](https://github.com/danny-avila/LibreChat/pull/6442)
+- 🏃‍♂️ refactor: More Agent Context Improvements during Run by **@danny-avila** in [#6477](https://github.com/danny-avila/LibreChat/pull/6477)
+- 🔃 refactor: Allow streaming for `o1` models by **@danny-avila** in [#6509](https://github.com/danny-avila/LibreChat/pull/6509)
+- 🔧 chore: `Vite` Plugin Upgrades & Config Optimizations by **@rubentalstra** in [#6547](https://github.com/danny-avila/LibreChat/pull/6547)
+- 🔧 refactor: Consolidate Logging, Model Selection & Actions Optimizations, Minor Fixes by **@danny-avila** in [#6553](https://github.com/danny-avila/LibreChat/pull/6553)
+- 🎨 style: Address Minor UI Refresh Issues by **@berry-13** in [#6552](https://github.com/danny-avila/LibreChat/pull/6552)
+- 🔧 refactor: Enhance Model & Endpoint Configurations with Global Indicators 🌍 by **@berry-13** in [#6578](https://github.com/danny-avila/LibreChat/pull/6578)
+- 💬 style: Chat UI, Greeting, and Message adjustments by **@berry-13** in [#6612](https://github.com/danny-avila/LibreChat/pull/6612)
+- ⚡ refactor: DocumentDB Compatibility for Balance Updates by **@danny-avila** in [#6673](https://github.com/danny-avila/LibreChat/pull/6673)
+- 🧹 chore: Update ESLint rules for React hooks by **@rubentalstra** in [#6685](https://github.com/danny-avila/LibreChat/pull/6685)
+- 🪙 chore: Update Gemini Pricing by **@RedwindA** in [#6731](https://github.com/danny-avila/LibreChat/pull/6731)
+- 🪺 refactor: Nest Permission fields for Roles by **@rubentalstra** in [#6487](https://github.com/danny-avila/LibreChat/pull/6487)
+- 📦 chore: Update `caniuse-lite` dependency to version 1.0.30001706 by **@rubentalstra** in [#6482](https://github.com/danny-avila/LibreChat/pull/6482)
+- ⚙️ refactor: OAuth Flow Signal, Type Safety, Tool Progress & Updated Packages by **@danny-avila** in [#6752](https://github.com/danny-avila/LibreChat/pull/6752)
+- 📦 chore: bump vite from 6.2.3 to 6.2.5 by **@dependabot[bot]** in [#6745](https://github.com/danny-avila/LibreChat/pull/6745)
+- 💾 chore: Enhance Local Storage Handling and Update MCP SDK by **@danny-avila** in [#6809](https://github.com/danny-avila/LibreChat/pull/6809)
+- 🤖 refactor: Improve Agents Memory Usage, Bump Keyv, Grok 3 by **@danny-avila** in [#6850](https://github.com/danny-avila/LibreChat/pull/6850)
+- 💾 refactor: Enhance Memory In Image Encodings & Client Disposal by **@danny-avila** in [#6852](https://github.com/danny-avila/LibreChat/pull/6852)
+- 🔁 refactor: Token Event Handler and Standardize `maxTokens` Key by **@danny-avila** in [#6886](https://github.com/danny-avila/LibreChat/pull/6886)
+- 🔍 refactor: Search & Message Retrieval by **@berry-13** in [#6903](https://github.com/danny-avila/LibreChat/pull/6903)
+- 🎨 style: standardize dropdown styling & fix z-Index layering by **@berry-13** in [#6939](https://github.com/danny-avila/LibreChat/pull/6939)
+- 📙 docs: CONTRIBUTING.md by **@dblock** in [#6831](https://github.com/danny-avila/LibreChat/pull/6831)
+- 🧭 refactor: Modernize Nav/Header by **@danny-avila** in [#7094](https://github.com/danny-avila/LibreChat/pull/7094)
+- 🪶 refactor: Chat Input Focus for Conversation Navigations & ChatForm Optimizations by **@danny-avila** in [#7100](https://github.com/danny-avila/LibreChat/pull/7100)
+- 🔃 refactor: Streamline Navigation, Message Loading UX by **@danny-avila** in [#7118](https://github.com/danny-avila/LibreChat/pull/7118)
+- 📜 docs: Unreleased changelog by **@github-actions[bot]** in [#6265](https://github.com/danny-avila/LibreChat/pull/6265)
+
+
+
+[See full release details][release-v0.7.8-rc1]
+
+[release-v0.7.8-rc1]: https://github.com/danny-avila/LibreChat/releases/tag/v0.7.8-rc1
+
+---
--- a/15
+++ b/15
@@ -1,9 +1,18 @@
-# v0.7.7-rc1
+# v0.7.9

 # Base node image
 FROM node:20-alpine AS node

-RUN apk --no-cache add curl
+# Install jemalloc
+RUN apk add --no-cache jemalloc
+RUN apk add --no-cache python3 py3-pip uv
+
+# Set environment variable to use jemalloc
+ENV LD_PRELOAD=/usr/lib/libjemalloc.so.2
+
+# Add `uv` for extended MCP support
+COPY --from=ghcr.io/astral-sh/uv:0.6.13 /uv /uvx /bin/
+RUN uv --version

 RUN mkdir -p /app && chown node:node /app
 WORKDIR /app
@@ -38,4 +47,4 @@ CMD ["npm", "run", "backend"]
 # WORKDIR /usr/share/nginx/html
 # COPY --from=node /app/client/dist /usr/share/nginx/html
 # COPY client/nginx.conf /etc/nginx/conf.d/default.conf
-# ENTRYPOINT ["nginx", "-g", "daemon off;"]
+# ENTRYPOINT ["nginx", "-g", "daemon off;"]
--- a/Dockerfile.multi
+++ b/Dockerfile.multi
@@ -1,8 +1,12 @@
 # Dockerfile.multi
-# v0.7.7-rc1
+# v0.7.9

 # Base for all builds
 FROM node:20-alpine AS base-min
+# Install jemalloc
+RUN apk add --no-cache jemalloc
+# Set environment variable to use jemalloc
+ENV LD_PRELOAD=/usr/lib/libjemalloc.so.2
 WORKDIR /app
 RUN apk --no-cache add curl
 RUN npm config set fetch-retry-maxtimeout 600000 && \
@@ -10,7 +14,8 @@ RUN npm config set fetch-retry-maxtimeout 600000 && \
    npm config set fetch-retry-mintimeout 15000
 COPY package*.json ./
 COPY packages/data-provider/package*.json ./packages/data-provider/
-COPY packages/mcp/package*.json ./packages/mcp/
+COPY packages/api/package*.json ./packages/api/
+COPY packages/data-schemas/package*.json ./packages/data-schemas/
 COPY client/package*.json ./client/
 COPY api/package*.json ./api/

@@ -19,19 +24,27 @@ FROM base-min AS base
 WORKDIR /app
 RUN npm ci

-# Build data-provider
+# Build `data-provider` package
 FROM base AS data-provider-build
 WORKDIR /app/packages/data-provider
 COPY packages/data-provider ./
 RUN npm run build

-# Build mcp package
-FROM base AS mcp-build
-WORKDIR /app/packages/mcp
-COPY packages/mcp ./
+# Build `data-schemas` package
+FROM base AS data-schemas-build
+WORKDIR /app/packages/data-schemas
+COPY packages/data-schemas ./
 COPY --from=data-provider-build /app/packages/data-provider/dist /app/packages/data-provider/dist
 RUN npm run build

+# Build `api` package
+FROM base AS api-package-build
+WORKDIR /app/packages/api
+COPY packages/api ./
+COPY --from=data-provider-build /app/packages/data-provider/dist /app/packages/data-provider/dist
+COPY --from=data-schemas-build /app/packages/data-schemas/dist /app/packages/data-schemas/dist
+RUN npm run build
+
 # Client build
 FROM base AS client-build
 WORKDIR /app/client
@@ -42,15 +55,19 @@ RUN npm run build

 # API setup (including client dist)
 FROM base-min AS api-build
+# Add `uv` for extended MCP support
+COPY --from=ghcr.io/astral-sh/uv:0.6.13 /uv /uvx /bin/
+RUN uv --version
 WORKDIR /app
 # Install only production deps
 RUN npm ci --omit=dev
 COPY api ./api
 COPY config ./config
 COPY --from=data-provider-build /app/packages/data-provider/dist ./packages/data-provider/dist
-COPY --from=mcp-build /app/packages/mcp/dist ./packages/mcp/dist
+COPY --from=data-schemas-build /app/packages/data-schemas/dist ./packages/data-schemas/dist
+COPY --from=api-package-build /app/packages/api/dist ./packages/api/dist
 COPY --from=client-build /app/client/dist ./client/dist
 WORKDIR /app/api
 EXPOSE 3080
 ENV HOST=0.0.0.0
-CMD ["node", "server/index.js"]
+CMD ["node", "server/index.js"]
--- a/README.md
+++ b/README.md
@@ -52,7 +52,7 @@
 - 🖥️ **UI & Experience** inspired by ChatGPT with enhanced design and features

 - 🤖 **AI Model Selection**:  
-  - Anthropic (Claude), AWS Bedrock, OpenAI, Azure OpenAI, Google, Vertex AI, OpenAI Assistants API (incl. Azure)
+  - Anthropic (Claude), AWS Bedrock, OpenAI, Azure OpenAI, Google, Vertex AI, OpenAI Responses API (incl. Azure)
  - [Custom Endpoints](https://www.librechat.ai/docs/quick_start/custom_endpoints): Use any OpenAI-compatible API with LibreChat, no proxy required
  - Compatible with [Local & Remote AI Providers](https://www.librechat.ai/docs/configuration/librechat_yaml/ai_endpoints):
    - Ollama, groq, Cohere, Mistral AI, Apple MLX, koboldcpp, together.ai,
@@ -66,14 +66,23 @@
 - 🔦 **Agents & Tools Integration**:  
  - **[LibreChat Agents](https://www.librechat.ai/docs/features/agents)**:
    - No-Code Custom Assistants: Build specialized, AI-driven helpers without coding  
-    - Flexible & Extensible: Attach tools like DALL-E-3, file search, code execution, and more  
-    - Compatible with Custom Endpoints, OpenAI, Azure, Anthropic, AWS Bedrock, and more
+    - Flexible & Extensible: Use MCP Servers, tools, file search, code execution, and more  
+    - Compatible with Custom Endpoints, OpenAI, Azure, Anthropic, AWS Bedrock, Google, Vertex AI, Responses API, and more
    - [Model Context Protocol (MCP) Support](https://modelcontextprotocol.io/clients#librechat) for Tools
-  - Use LibreChat Agents and OpenAI Assistants with Files, Code Interpreter, Tools, and API Actions
+
+- 🔍 **Web Search**:  
+  - Search the internet and retrieve relevant information to enhance your AI context
+  - Combines search providers, content scrapers, and result rerankers for optimal results
+  - **[Learn More →](https://www.librechat.ai/docs/features/web_search)**

 - 🪄 **Generative UI with Code Artifacts**:  
  - [Code Artifacts](https://youtu.be/GfTj7O4gmd0?si=WJbdnemZpJzBrJo3) allow creation of React, HTML, and Mermaid diagrams directly in chat

+- 🎨 **Image Generation & Editing**
+  - Text-to-image and image-to-image with [GPT-Image-1](https://www.librechat.ai/docs/features/image_gen#1--openai-image-tools-recommended)
+  - Text-to-image with [DALL-E (3/2)](https://www.librechat.ai/docs/features/image_gen#2--dalle-legacy), [Stable Diffusion](https://www.librechat.ai/docs/features/image_gen#3--stable-diffusion-local), [Flux](https://www.librechat.ai/docs/features/image_gen#4--flux), or any [MCP server](https://www.librechat.ai/docs/features/image_gen#5--model-context-protocol-mcp)
+  - Produce stunning visuals from prompts or refine existing images with a single instruction
+
 - 💾 **Presets & Context Management**:  
  - Create, Save, & Share Custom Presets  
  - Switch between AI Endpoints and Presets mid-chat
@@ -81,7 +90,7 @@
  - [Fork Messages & Conversations](https://www.librechat.ai/docs/features/fork) for Advanced Context control

 - 💬 **Multimodal & File Interactions**:  
-  - Upload and analyze images with Claude 3, GPT-4o, o1, Llama-Vision, and Gemini 📸  
+  - Upload and analyze images with Claude 3, GPT-4.5, GPT-4o, o1, Llama-Vision, and Gemini 📸  
  - Chat with Files using Custom Endpoints, OpenAI, Azure, Anthropic, AWS Bedrock, & Google 🗃️

 - 🌎 **Multilingual UI**:  
@@ -140,8 +149,8 @@ Click on the thumbnail to open the video☝️

 **Other:**
  - **Website:** [librechat.ai](https://librechat.ai)
-  - **Documentation:** [docs.librechat.ai](https://docs.librechat.ai)
-  - **Blog:** [blog.librechat.ai](https://blog.librechat.ai)
+  - **Documentation:** [librechat.ai/docs](https://librechat.ai/docs)
+  - **Blog:** [librechat.ai/blog](https://librechat.ai/blog)

 ---

@@ -197,6 +206,6 @@ We thank [Locize](https://locize.com) for their translation management tools tha

 <p align="center">
  <a href="https://locize.com" target="_blank" rel="noopener noreferrer">
-    <img src="https://locize.com/img/locize_color.svg" alt="Locize Logo" height="50">
+    <img src="https://github.com/user-attachments/assets/d6b70894-6064-475e-bb65-92a9e23e0077" alt="Locize Logo" height="50">
  </a>
 </p>
--- a/api/app/clients/AnthropicClient.js
+++ b/api/app/clients/AnthropicClient.js
@@ -2,12 +2,15 @@ const Anthropic = require('@anthropic-ai/sdk');
 const { HttpsProxyAgent } = require('https-proxy-agent');
 const {
  Constants,
+  ErrorTypes,
  EModelEndpoint,
+  parseTextParts,
  anthropicSettings,
  getResponseSender,
  validateVisionModel,
 } = require('librechat-data-provider');
-const { encodeAndFormat } = require('~/server/services/Files/images/encode');
+const { SplitStreamHandler: _Handler } = require('@librechat/agents');
+const { Tokenizer, createFetch, createStreamEventHandlers } = require('@librechat/api');
 const {
  truncateText,
  formatMessage,
@@ -16,9 +19,14 @@ const {
  parseParamFromPrompt,
  createContextHandlers,
 } = require('./prompts');
+const {
+  getClaudeHeaders,
+  configureReasoning,
+  checkPromptCacheSupport,
+} = require('~/server/services/Endpoints/anthropic/helpers');
 const { getModelMaxTokens, getModelMaxOutputTokens, matchModelName } = require('~/utils');
 const { spendTokens, spendStructuredTokens } = require('~/models/spendTokens');
-const Tokenizer = require('~/server/services/Tokenizer');
+const { encodeAndFormat } = require('~/server/services/Files/images/encode');
 const { sleep } = require('~/server/utils');
 const BaseClient = require('./BaseClient');
 const { logger } = require('~/config');
@@ -26,6 +34,15 @@ const { logger } = require('~/config');
 const HUMAN_PROMPT = '\n\nHuman:';
 const AI_PROMPT = '\n\nAssistant:';

+class SplitStreamHandler extends _Handler {
+  getDeltaContent(chunk) {
+    return (chunk?.delta?.text ?? chunk?.completion) || '';
+  }
+  getReasoningDelta(chunk) {
+    return chunk?.delta?.thinking || '';
+  }
+}
+
 /** Helper function to introduce a delay before retrying */
 function delayBeforeRetry(attempts, baseDelay = 1000) {
  return new Promise((resolve) => setTimeout(resolve, baseDelay * attempts));
@@ -52,13 +69,10 @@ class AnthropicClient extends BaseClient {
    this.message_delta;
    /** Whether the model is part of the Claude 3 Family
     * @type {boolean} */
-    this.isClaude3;
+    this.isClaudeLatest;
    /** Whether to use Messages API or Completions API
     * @type {boolean} */
    this.useMessages;
-    /** Whether or not the model is limited to the legacy amount of output tokens
-     * @type {boolean} */
-    this.isLegacyOutput;
    /** Whether or not the model supports Prompt Caching
     * @type {boolean} */
    this.supportsCacheControl;
@@ -68,6 +82,8 @@ class AnthropicClient extends BaseClient {
    /** The key for the usage object's output tokens
     * @type {string} */
    this.outputTokensKey = 'output_tokens';
+    /** @type {SplitStreamHandler | undefined} */
+    this.streamHandler;
  }

  setOptions(options) {
@@ -96,20 +112,25 @@ class AnthropicClient extends BaseClient {
    );

    const modelMatch = matchModelName(this.modelOptions.model, EModelEndpoint.anthropic);
-    this.isClaude3 = modelMatch.includes('claude-3');
-    this.isLegacyOutput = !modelMatch.includes('claude-3-5-sonnet');
-    this.supportsCacheControl =
-      this.options.promptCache && this.checkPromptCacheSupport(modelMatch);
+    this.isClaudeLatest =
+      /claude-[3-9]/.test(modelMatch) || /claude-(?:sonnet|opus|haiku)-[4-9]/.test(modelMatch);
+    const isLegacyOutput = !(
+      /claude-3[-.]5-sonnet/.test(modelMatch) ||
+      /claude-3[-.]7/.test(modelMatch) ||
+      /claude-(?:sonnet|opus|haiku)-[4-9]/.test(modelMatch) ||
+      /claude-[4-9]/.test(modelMatch)
+    );
+    this.supportsCacheControl = this.options.promptCache && checkPromptCacheSupport(modelMatch);

    if (
-      this.isLegacyOutput &&
+      isLegacyOutput &&
      this.modelOptions.maxOutputTokens &&
      this.modelOptions.maxOutputTokens > legacy.maxOutputTokens.default
    ) {
      this.modelOptions.maxOutputTokens = legacy.maxOutputTokens.default;
    }

-    this.useMessages = this.isClaude3 || !!this.options.attachments;
+    this.useMessages = this.isClaudeLatest || !!this.options.attachments;

    this.defaultVisionModel = this.options.visionModel ?? 'claude-3-sonnet-20240229';
    this.options.attachments?.then((attachments) => this.checkVisionRequest(attachments));
@@ -125,16 +146,21 @@ class AnthropicClient extends BaseClient {
        this.options.endpointType ?? this.options.endpoint,
        this.options.endpointTokenConfig,
      ) ??
-      1500;
+      anthropicSettings.maxOutputTokens.reset(this.modelOptions.model);
    this.maxPromptTokens =
      this.options.maxPromptTokens || this.maxContextTokens - this.maxResponseTokens;

-    if (this.maxPromptTokens + this.maxResponseTokens > this.maxContextTokens) {
-      throw new Error(
-        `maxPromptTokens + maxOutputTokens (${this.maxPromptTokens} + ${this.maxResponseTokens} = ${
-          this.maxPromptTokens + this.maxResponseTokens
-        }) must be less than or equal to maxContextTokens (${this.maxContextTokens})`,
-      );
+    const reservedTokens = this.maxPromptTokens + this.maxResponseTokens;
+    if (reservedTokens > this.maxContextTokens) {
+      const info = `Total Possible Tokens + Max Output Tokens must be less than or equal to Max Context Tokens: ${this.maxPromptTokens} (total possible output) + ${this.maxResponseTokens} (max output) = ${reservedTokens}/${this.maxContextTokens} (max context)`;
+      const errorMessage = `{ "type": "${ErrorTypes.INPUT_LENGTH}", "info": "${info}" }`;
+      logger.warn(info);
+      throw new Error(errorMessage);
+    } else if (this.maxResponseTokens === this.maxContextTokens) {
+      const info = `Max Output Tokens must be less than Max Context Tokens: ${this.maxResponseTokens} (max output) = ${this.maxContextTokens} (max context)`;
+      const errorMessage = `{ "type": "${ErrorTypes.INPUT_LENGTH}", "info": "${info}" }`;
+      logger.warn(info);
+      throw new Error(errorMessage);
    }

    this.sender =
@@ -159,30 +185,25 @@ class AnthropicClient extends BaseClient {
  getClient(requestOptions) {
    /** @type {Anthropic.ClientOptions} */
    const options = {
-      fetch: this.fetch,
+      fetch: createFetch({
+        directEndpoint: this.options.directEndpoint,
+        reverseProxyUrl: this.options.reverseProxyUrl,
+      }),
      apiKey: this.apiKey,
+      fetchOptions: {},
    };

    if (this.options.proxy) {
-      options.httpAgent = new HttpsProxyAgent(this.options.proxy);
+      options.fetchOptions.agent = new HttpsProxyAgent(this.options.proxy);
    }

    if (this.options.reverseProxyUrl) {
      options.baseURL = this.options.reverseProxyUrl;
    }

-    if (
-      this.supportsCacheControl &&
-      requestOptions?.model &&
-      requestOptions.model.includes('claude-3-5-sonnet')
-    ) {
-      options.defaultHeaders = {
-        'anthropic-beta': 'max-tokens-3-5-sonnet-2024-07-15,prompt-caching-2024-07-31',
-      };
-    } else if (this.supportsCacheControl) {
-      options.defaultHeaders = {
-        'anthropic-beta': 'prompt-caching-2024-07-31',
-      };
+    const headers = getClaudeHeaders(requestOptions?.model, this.supportsCacheControl);
+    if (headers) {
+      options.defaultHeaders = headers;
    }

    return new Anthropic(options);
@@ -376,13 +397,13 @@ class AnthropicClient extends BaseClient {
    const formattedMessages = orderedMessages.map((message, i) => {
      const formattedMessage = this.useMessages
        ? formatMessage({
-          message,
-          endpoint: EModelEndpoint.anthropic,
-        })
+            message,
+            endpoint: EModelEndpoint.anthropic,
+          })
        : {
-          author: message.isCreatedByUser ? this.userLabel : this.assistantLabel,
-          content: message?.content ?? message.text,
-        };
+            author: message.isCreatedByUser ? this.userLabel : this.assistantLabel,
+            content: message?.content ?? message.text,
+          };

      const needsTokenCount = this.contextStrategy && !orderedMessages[i].tokenCount;
      /* If tokens were never counted, or, is a Vision request and the message has files, count again */
@@ -398,6 +419,9 @@ class AnthropicClient extends BaseClient {
            this.contextHandlers?.processFile(file);
            continue;
          }
+          if (file.metadata?.fileIdentifier) {
+            continue;
+          }

          orderedMessages[i].tokenCount += this.calculateImageTokenCost({
            width: file.width,
@@ -631,7 +655,10 @@ class AnthropicClient extends BaseClient {
      );
    };

-    if (this.modelOptions.model.includes('claude-3')) {
+    if (
+      /claude-[3-9]/.test(this.modelOptions.model) ||
+      /claude-(?:sonnet|opus|haiku)-[4-9]/.test(this.modelOptions.model)
+    ) {
      await buildMessagesPayload();
      processTokens();
      return {
@@ -657,7 +684,7 @@ class AnthropicClient extends BaseClient {
  }

  getCompletion() {
-    logger.debug('AnthropicClient doesn\'t use getCompletion (all handled in sendCompletion)');
+    logger.debug("AnthropicClient doesn't use getCompletion (all handled in sendCompletion)");
  }

  /**
@@ -668,29 +695,41 @@ class AnthropicClient extends BaseClient {
   * @returns {Promise<Anthropic.default.Message | Anthropic.default.Completion>} The response from the Anthropic client.
   */
  async createResponse(client, options, useMessages) {
-    return useMessages ?? this.useMessages
+    return (useMessages ?? this.useMessages)
      ? await client.messages.create(options)
      : await client.completions.create(options);
  }

+  getMessageMapMethod() {
+    /**
+     * @param {TMessage} msg
+     */
+    return (msg) => {
+      if (msg.text != null && msg.text && msg.text.startsWith(':::thinking')) {
+        msg.text = msg.text.replace(/:::thinking.*?:::/gs, '').trim();
+      } else if (msg.content != null) {
+        msg.text = parseTextParts(msg.content, true);
+        delete msg.content;
+      }
+
+      return msg;
+    };
+  }
+
  /**
-   * @param {string} modelName
-   * @returns {boolean}
+   * @param {string[]} [intermediateReply]
+   * @returns {string}
   */
-  checkPromptCacheSupport(modelName) {
-    const modelMatch = matchModelName(modelName, EModelEndpoint.anthropic);
-    if (modelMatch.includes('claude-3-5-sonnet-latest')) {
-      return false;
+  getStreamText(intermediateReply) {
+    if (!this.streamHandler) {
+      return intermediateReply?.join('') ?? '';
    }
-    if (
-      modelMatch === 'claude-3-5-sonnet' ||
-      modelMatch === 'claude-3-5-haiku' ||
-      modelMatch === 'claude-3-haiku' ||
-      modelMatch === 'claude-3-opus'
-    ) {
-      return true;
-    }
-    return false;
+
+    const reasoningText = this.streamHandler.reasoningTokens.join('');
+
+    const reasoningBlock = reasoningText.length > 0 ? `:::thinking\n${reasoningText}\n:::\n` : '';
+
+    return `${reasoningBlock}${this.streamHandler.tokens.join('')}`;
  }

  async sendCompletion(payload, { onProgress, abortController }) {
@@ -710,7 +749,6 @@ class AnthropicClient extends BaseClient {
      user_id: this.user,
    };

-    let text = '';
    const {
      stream,
      model,
@@ -721,22 +759,34 @@ class AnthropicClient extends BaseClient {
      topK: top_k,
    } = this.modelOptions;

-    const requestOptions = {
+    let requestOptions = {
      model,
      stream: stream || true,
      stop_sequences,
      temperature,
      metadata,
-      top_p,
-      top_k,
    };

    if (this.useMessages) {
      requestOptions.messages = payload;
-      requestOptions.max_tokens = maxOutputTokens || legacy.maxOutputTokens.default;
+      requestOptions.max_tokens =
+        maxOutputTokens || anthropicSettings.maxOutputTokens.reset(requestOptions.model);
    } else {
      requestOptions.prompt = payload;
-      requestOptions.max_tokens_to_sample = maxOutputTokens || 1500;
+      requestOptions.max_tokens_to_sample = maxOutputTokens || legacy.maxOutputTokens.default;
+    }
+
+    requestOptions = configureReasoning(requestOptions, {
+      thinking: this.options.thinking,
+      thinkingBudget: this.options.thinkingBudget,
+    });
+
+    if (!/claude-3[-.]7/.test(model)) {
+      requestOptions.top_p = top_p;
+      requestOptions.top_k = top_k;
+    } else if (requestOptions.thinking == null) {
+      requestOptions.topP = top_p;
+      requestOptions.topK = top_k;
    }

    if (this.systemMessage && this.supportsCacheControl === true) {
@@ -756,13 +806,14 @@ class AnthropicClient extends BaseClient {
    }

    logger.debug('[AnthropicClient]', { ...requestOptions });
+    const handlers = createStreamEventHandlers(this.options.res);
+    this.streamHandler = new SplitStreamHandler({
+      accumulate: true,
+      runId: this.responseMessageId,
+      handlers,
+    });

-    const handleChunk = (currentChunk) => {
-      if (currentChunk) {
-        text += currentChunk;
-        onProgress(currentChunk);
-      }
-    };
+    let intermediateReply = this.streamHandler.tokens;

    const maxRetries = 3;
    const streamRate = this.options.streamRate ?? Constants.DEFAULT_STREAM_RATE;
@@ -783,22 +834,15 @@ class AnthropicClient extends BaseClient {
          });

          for await (const completion of response) {
-            // Handle each completion as before
            const type = completion?.type ?? '';
            if (tokenEventTypes.has(type)) {
              logger.debug(`[AnthropicClient] ${type}`, completion);
              this[type] = completion;
            }
-            if (completion?.delta?.text) {
-              handleChunk(completion.delta.text);
-            } else if (completion.completion) {
-              handleChunk(completion.completion);
-            }
-
+            this.streamHandler.handle(completion);
            await sleep(streamRate);
          }

-          // Successful processing, exit loop
          break;
        } catch (error) {
          attempts += 1;
@@ -808,6 +852,10 @@ class AnthropicClient extends BaseClient {

          if (attempts < maxRetries) {
            await delayBeforeRetry(attempts, 350);
+          } else if (this.streamHandler && this.streamHandler.reasoningTokens.length) {
+            return this.getStreamText();
+          } else if (intermediateReply.length > 0) {
+            return this.getStreamText(intermediateReply);
          } else {
            throw new Error(`Operation failed after ${maxRetries} attempts: ${error.message}`);
          }
@@ -823,8 +871,7 @@ class AnthropicClient extends BaseClient {
    }

    await processResponse.bind(this)();
-
-    return text.trim();
+    return this.getStreamText(intermediateReply);
  }

  getSaveOptions() {
@@ -834,6 +881,8 @@ class AnthropicClient extends BaseClient {
      promptPrefix: this.options.promptPrefix,
      modelLabel: this.options.modelLabel,
      promptCache: this.options.promptCache,
+      thinking: this.options.thinking,
+      thinkingBudget: this.options.thinkingBudget,
      resendFiles: this.options.resendFiles,
      iconURL: this.options.iconURL,
      greeting: this.options.greeting,
@@ -843,7 +892,7 @@ class AnthropicClient extends BaseClient {
  }

  getBuildMessagesOptions() {
-    logger.debug('AnthropicClient doesn\'t use getBuildMessagesOptions');
+    logger.debug("AnthropicClient doesn't use getBuildMessagesOptions");
  }

  getEncoding() {
--- a/api/app/clients/BaseClient.js
+++ b/api/app/clients/BaseClient.js
@@ -5,13 +5,14 @@ const {
  isAgentsEndpoint,
  isParamEndpoint,
  EModelEndpoint,
+  ContentTypes,
+  excludedKeys,
  ErrorTypes,
  Constants,
 } = require('librechat-data-provider');
-const { getMessages, saveMessage, updateMessage, saveConvo } = require('~/models');
-const { addSpaceIfNeeded, isEnabled } = require('~/server/utils');
+const { getMessages, saveMessage, updateMessage, saveConvo, getConvo } = require('~/models');
+const { checkBalance } = require('~/models/balanceMethods');
 const { truncateToolCallOutputs } = require('./prompts');
-const checkBalance = require('~/models/checkBalance');
 const { getFiles } = require('~/models/File');
 const TextStream = require('./TextStream');
 const { logger } = require('~/config');
@@ -26,15 +27,10 @@ class BaseClient {
      month: 'long',
      day: 'numeric',
    });
-    this.fetch = this.fetch.bind(this);
    /** @type {boolean} */
    this.skipSaveConvo = false;
    /** @type {boolean} */
    this.skipSaveUserMessage = false;
-    /** @type {ClientDatabaseSavePromise} */
-    this.userMessagePromise;
-    /** @type {ClientDatabaseSavePromise} */
-    this.responsePromise;
    /** @type {string} */
    this.user;
    /** @type {string} */
@@ -55,6 +51,10 @@ class BaseClient {
     * Flag to determine if the client re-submitted the latest assistant message.
     * @type {boolean | undefined} */
    this.continued;
+    /**
+     * Flag to determine if the client has already fetched the conversation while saving new messages.
+     * @type {boolean | undefined} */
+    this.fetchedConvo;
    /** @type {TMessage[]} */
    this.currentMessages = [];
    /** @type {import('librechat-data-provider').VisionModes | undefined} */
@@ -62,15 +62,15 @@ class BaseClient {
  }

  setOptions() {
-    throw new Error('Method \'setOptions\' must be implemented.');
+    throw new Error("Method 'setOptions' must be implemented.");
  }

  async getCompletion() {
-    throw new Error('Method \'getCompletion\' must be implemented.');
+    throw new Error("Method 'getCompletion' must be implemented.");
  }

  async sendCompletion() {
-    throw new Error('Method \'sendCompletion\' must be implemented.');
+    throw new Error("Method 'sendCompletion' must be implemented.");
  }

  getSaveOptions() {
@@ -108,12 +108,15 @@ class BaseClient {
  /**
   * Abstract method to record token usage. Subclasses must implement this method.
   * If a correction to the token usage is needed, the method should return an object with the corrected token counts.
+   * Should only be used if `recordCollectedUsage` was not used instead.
+   * @param {string} [model]
   * @param {number} promptTokens
   * @param {number} completionTokens
   * @returns {Promise<void>}
   */
-  async recordTokenUsage({ promptTokens, completionTokens }) {
+  async recordTokenUsage({ model, promptTokens, completionTokens }) {
    logger.debug('[BaseClient] `recordTokenUsage` not implemented.', {
+      model,
      promptTokens,
      completionTokens,
    });
@@ -197,6 +200,10 @@ class BaseClient {
      this.currentMessages[this.currentMessages.length - 1].messageId = head;
    }

+    if (opts.isRegenerate && responseMessageId.endsWith('_')) {
+      responseMessageId = crypto.randomUUID();
+    }
+
    this.responseMessageId = responseMessageId;

    return {
@@ -236,11 +243,11 @@ class BaseClient {
    const userMessage = opts.isEdited
      ? this.currentMessages[this.currentMessages.length - 2]
      : this.createUserMessage({
-        messageId: userMessageId,
-        parentMessageId,
-        conversationId,
-        text: message,
-      });
+          messageId: userMessageId,
+          parentMessageId,
+          conversationId,
+          text: message,
+        });

    if (typeof opts?.getReqData === 'function') {
      opts.getReqData({
@@ -360,17 +367,14 @@ class BaseClient {
   *  context: TMessage[],
   *  remainingContextTokens: number,
   *  messagesToRefine: TMessage[],
-   *  summaryIndex: number,
-   * }>} An object with four properties: `context`, `summaryIndex`, `remainingContextTokens`, and `messagesToRefine`.
+   * }>} An object with three properties: `context`, `remainingContextTokens`, and `messagesToRefine`.
   *    `context` is an array of messages that fit within the token limit.
-   *    `summaryIndex` is the index of the first message in the `messagesToRefine` array.
   *    `remainingContextTokens` is the number of tokens remaining within the limit after adding the messages to the context.
   *    `messagesToRefine` is an array of messages that were not added to the context because they would have exceeded the token limit.
   */
  async getMessagesWithinTokenLimit({ messages: _messages, maxContextTokens, instructions }) {
    // Every reply is primed with <|start|>assistant<|message|>, so we
    // start with 3 tokens for the label after all messages have been counted.
-    let summaryIndex = -1;
    let currentTokenCount = 3;
    const instructionsTokenCount = instructions?.tokenCount ?? 0;
    let remainingContextTokens =
@@ -403,14 +407,12 @@ class BaseClient {
    }

    const prunedMemory = messages;
-    summaryIndex = prunedMemory.length - 1;
    remainingContextTokens -= currentTokenCount;

    return {
      context: context.reverse(),
      remainingContextTokens,
      messagesToRefine: prunedMemory,
-      summaryIndex,
    };
  }

@@ -453,7 +455,7 @@ class BaseClient {

    let orderedWithInstructions = this.addInstructions(orderedMessages, instructions);

-    let { context, remainingContextTokens, messagesToRefine, summaryIndex } =
+    let { context, remainingContextTokens, messagesToRefine } =
      await this.getMessagesWithinTokenLimit({
        messages: orderedWithInstructions,
        instructions,
@@ -523,7 +525,7 @@ class BaseClient {
    }

    // Make sure to only continue summarization logic if the summary message was generated
-    shouldSummarize = summaryMessage && shouldSummarize;
+    shouldSummarize = summaryMessage != null && shouldSummarize === true;

    logger.debug('[BaseClient] Context Count (2/2)', {
      remainingContextTokens,
@@ -533,17 +535,18 @@ class BaseClient {
    /** @type {Record<string, number> | undefined} */
    let tokenCountMap;
    if (buildTokenMap) {
-      tokenCountMap = orderedWithInstructions.reduce((map, message, index) => {
+      const currentPayload = shouldSummarize ? orderedWithInstructions : context;
+      tokenCountMap = currentPayload.reduce((map, message, index) => {
        const { messageId } = message;
        if (!messageId) {
          return map;
        }

-        if (shouldSummarize && index === summaryIndex && !usePrevSummary) {
+        if (shouldSummarize && index === messagesToRefine.length - 1 && !usePrevSummary) {
          map.summaryMessage = { ...summaryMessage, messageId, tokenCount: summaryTokenCount };
        }

-        map[messageId] = orderedWithInstructions[index].tokenCount;
+        map[messageId] = currentPayload[index].tokenCount;
        return map;
      }, {});
    }
@@ -562,6 +565,8 @@ class BaseClient {
  }

  async sendMessage(message, opts = {}) {
+    /** @type {Promise<TMessage>} */
+    let userMessagePromise;
    const { user, head, isEdited, conversationId, responseMessageId, saveOptions, userMessage } =
      await this.handleStartMethods(message, opts);

@@ -573,7 +578,7 @@ class BaseClient {
      });
    }

-    const { generation = '' } = opts;
+    const { editedContent } = opts;

    // It's not necessary to push to currentMessages
    // depending on subclass implementation of handling messages
@@ -588,11 +593,21 @@ class BaseClient {
          isCreatedByUser: false,
          model: this.modelOptions?.model ?? this.model,
          sender: this.sender,
-          text: generation,
        };
        this.currentMessages.push(userMessage, latestMessage);
-      } else {
-        latestMessage.text = generation;
+      } else if (editedContent != null) {
+        // Handle editedContent for content parts
+        if (editedContent && latestMessage.content && Array.isArray(latestMessage.content)) {
+          const { index, text, type } = editedContent;
+          if (index >= 0 && index < latestMessage.content.length) {
+            const contentPart = latestMessage.content[index];
+            if (type === ContentTypes.THINK && contentPart.type === ContentTypes.THINK) {
+              contentPart[ContentTypes.THINK] = text;
+            } else if (type === ContentTypes.TEXT && contentPart.type === ContentTypes.TEXT) {
+              contentPart[ContentTypes.TEXT] = text;
+            }
+          }
+        }
      }
      this.continued = true;
    } else {
@@ -623,17 +638,18 @@ class BaseClient {
    }

    if (!isEdited && !this.skipSaveUserMessage) {
-      this.userMessagePromise = this.saveMessageToDatabase(userMessage, saveOptions, user);
+      userMessagePromise = this.saveMessageToDatabase(userMessage, saveOptions, user);
      this.savedMessageIds.add(userMessage.messageId);
      if (typeof opts?.getReqData === 'function') {
        opts.getReqData({
-          userMessagePromise: this.userMessagePromise,
+          userMessagePromise,
        });
      }
    }

+    const balance = this.options.req?.app?.locals?.balance;
    if (
-      isEnabled(process.env.CHECK_BALANCE) &&
+      balance?.enabled &&
      supportsBalanceCheck[this.options.endpointType ?? this.options.endpoint]
    ) {
      await checkBalance({
@@ -652,7 +668,9 @@ class BaseClient {

    /** @type {string|string[]|undefined} */
    const completion = await this.sendCompletion(payload, opts);
-    this.abortController.requestCompleted = true;
+    if (this.abortController) {
+      this.abortController.requestCompleted = true;
+    }

    /** @type {TMessage} */
    const responseMessage = {
@@ -670,15 +688,32 @@ class BaseClient {
    };

    if (typeof completion === 'string') {
-      responseMessage.text = addSpaceIfNeeded(generation) + completion;
+      responseMessage.text = completion;
    } else if (
      Array.isArray(completion) &&
-      isParamEndpoint(this.options.endpoint, this.options.endpointType)
+      (this.clientName === EModelEndpoint.agents ||
+        isParamEndpoint(this.options.endpoint, this.options.endpointType))
    ) {
      responseMessage.text = '';
-      responseMessage.content = completion;
+
+      if (!opts.editedContent || this.currentMessages.length === 0) {
+        responseMessage.content = completion;
+      } else {
+        const latestMessage = this.currentMessages[this.currentMessages.length - 1];
+        if (!latestMessage?.content) {
+          responseMessage.content = completion;
+        } else {
+          const existingContent = [...latestMessage.content];
+          const { type: editedType } = opts.editedContent;
+          responseMessage.content = this.mergeEditedContent(
+            existingContent,
+            completion,
+            editedType,
+          );
+        }
+      }
    } else if (Array.isArray(completion)) {
-      responseMessage.text = addSpaceIfNeeded(generation) + completion.join('');
+      responseMessage.text = completion.join('');
    }

    if (
@@ -699,17 +734,27 @@ class BaseClient {
      if (usage != null && Number(usage[this.outputTokensKey]) > 0) {
        responseMessage.tokenCount = usage[this.outputTokensKey];
        completionTokens = responseMessage.tokenCount;
-        await this.updateUserMessageTokenCount({ usage, tokenCountMap, userMessage, opts });
+        await this.updateUserMessageTokenCount({
+          usage,
+          tokenCountMap,
+          userMessage,
+          userMessagePromise,
+          opts,
+        });
      } else {
        responseMessage.tokenCount = this.getTokenCountForResponse(responseMessage);
        completionTokens = responseMessage.tokenCount;
+        await this.recordTokenUsage({
+          usage,
+          promptTokens,
+          completionTokens,
+          model: responseMessage.model,
+        });
      }
-
-      await this.recordTokenUsage({ promptTokens, completionTokens, usage });
    }

-    if (this.userMessagePromise) {
-      await this.userMessagePromise;
+    if (userMessagePromise) {
+      await userMessagePromise;
    }

    if (this.artifactPromises) {
@@ -724,7 +769,11 @@ class BaseClient {
      }
    }

-    this.responsePromise = this.saveMessageToDatabase(responseMessage, saveOptions, user);
+    responseMessage.databasePromise = this.saveMessageToDatabase(
+      responseMessage,
+      saveOptions,
+      user,
+    );
    this.savedMessageIds.add(responseMessage.messageId);
    delete responseMessage.tokenCount;
    return responseMessage;
@@ -745,9 +794,16 @@ class BaseClient {
   * @param {StreamUsage} params.usage
   * @param {Record<string, number>} params.tokenCountMap
   * @param {TMessage} params.userMessage
+   * @param {Promise<TMessage>} params.userMessagePromise
   * @param {object} params.opts
   */
-  async updateUserMessageTokenCount({ usage, tokenCountMap, userMessage, opts }) {
+  async updateUserMessageTokenCount({
+    usage,
+    tokenCountMap,
+    userMessage,
+    userMessagePromise,
+    opts,
+  }) {
    /** @type {boolean} */
    const shouldUpdateCount =
      this.calculateCurrentTokenCount != null &&
@@ -772,7 +828,8 @@ class BaseClient {

    userMessage.tokenCount = userMessageTokenCount;
    /*
-      Note: `AskController` saves the user message, so we update the count of its `userMessage` reference
+      Note: `AgentController` saves the user message if not saved here
+      (noted by `savedMessageIds`), so we update the count of its `userMessage` reference
    */
    if (typeof opts?.getReqData === 'function') {
      opts.getReqData({
@@ -781,9 +838,10 @@ class BaseClient {
    }
    /*
      Note: we update the user message to be sure it gets the calculated token count;
-      though `AskController` saves the user message, EditController does not
+      though `AgentController` saves the user message if not saved here
+      (noted by `savedMessageIds`), EditController does not
    */
-    await this.userMessagePromise;
+    await userMessagePromise;
    await this.updateMessageInDatabase({
      messageId: userMessage.messageId,
      tokenCount: userMessageTokenCount,
@@ -849,7 +907,7 @@ class BaseClient {
    }

    const savedMessage = await saveMessage(
-      this.options.req,
+      this.options?.req,
      {
        ...message,
        endpoint: this.options.endpoint,
@@ -863,16 +921,40 @@ class BaseClient {
      return { message: savedMessage };
    }

-    const conversation = await saveConvo(
-      this.options.req,
-      {
-        conversationId: message.conversationId,
-        endpoint: this.options.endpoint,
-        endpointType: this.options.endpointType,
-        ...endpointOptions,
-      },
-      { context: 'api/app/clients/BaseClient.js - saveMessageToDatabase #saveConvo' },
-    );
+    const fieldsToKeep = {
+      conversationId: message.conversationId,
+      endpoint: this.options.endpoint,
+      endpointType: this.options.endpointType,
+      ...endpointOptions,
+    };
+
+    const existingConvo =
+      this.fetchedConvo === true
+        ? null
+        : await getConvo(this.options?.req?.user?.id, message.conversationId);
+
+    const unsetFields = {};
+    const exceptions = new Set(['spec', 'iconURL']);
+    if (existingConvo != null) {
+      this.fetchedConvo = true;
+      for (const key in existingConvo) {
+        if (!key) {
+          continue;
+        }
+        if (excludedKeys.has(key) && !exceptions.has(key)) {
+          continue;
+        }
+
+        if (endpointOptions?.[key] === undefined) {
+          unsetFields[key] = 1;
+        }
+      }
+    }
+
+    const conversation = await saveConvo(this.options?.req, fieldsToKeep, {
+      context: 'api/app/clients/BaseClient.js - saveMessageToDatabase #saveConvo',
+      unsetFields,
+    });

    return { message: savedMessage, conversation };
  }
@@ -993,11 +1075,17 @@ class BaseClient {
    const processValue = (value) => {
      if (Array.isArray(value)) {
        for (let item of value) {
-          if (!item || !item.type || item.type === 'image_url') {
+          if (
+            !item ||
+            !item.type ||
+            item.type === ContentTypes.THINK ||
+            item.type === ContentTypes.ERROR ||
+            item.type === ContentTypes.IMAGE_URL
+          ) {
            continue;
          }

-          if (item.type === 'tool_call' && item.tool_call != null) {
+          if (item.type === ContentTypes.TOOL_CALL && item.tool_call != null) {
            const toolName = item.tool_call?.name || '';
            if (toolName != null && toolName && typeof toolName === 'string') {
              numTokens += this.getTokenCount(toolName);
@@ -1043,6 +1131,50 @@ class BaseClient {
    return numTokens;
  }

+  /**
+   * Merges completion content with existing content when editing TEXT or THINK types
+   * @param {Array} existingContent - The existing content array
+   * @param {Array} newCompletion - The new completion content
+   * @param {string} editedType - The type of content being edited
+   * @returns {Array} The merged content array
+   */
+  mergeEditedContent(existingContent, newCompletion, editedType) {
+    if (!newCompletion.length) {
+      return existingContent.concat(newCompletion);
+    }
+
+    if (editedType !== ContentTypes.TEXT && editedType !== ContentTypes.THINK) {
+      return existingContent.concat(newCompletion);
+    }
+
+    const lastIndex = existingContent.length - 1;
+    const lastExisting = existingContent[lastIndex];
+    const firstNew = newCompletion[0];
+
+    if (lastExisting?.type !== firstNew?.type || firstNew?.type !== editedType) {
+      return existingContent.concat(newCompletion);
+    }
+
+    const mergedContent = [...existingContent];
+    if (editedType === ContentTypes.TEXT) {
+      mergedContent[lastIndex] = {
+        ...mergedContent[lastIndex],
+        [ContentTypes.TEXT]:
+          (mergedContent[lastIndex][ContentTypes.TEXT] || '') + (firstNew[ContentTypes.TEXT] || ''),
+      };
+    } else {
+      mergedContent[lastIndex] = {
+        ...mergedContent[lastIndex],
+        [ContentTypes.THINK]:
+          (mergedContent[lastIndex][ContentTypes.THINK] || '') +
+          (firstNew[ContentTypes.THINK] || ''),
+      };
+    }
+
+    // Add remaining completion items
+    return mergedContent.concat(newCompletion.slice(1));
+  }
+
  async sendPayload(payload, opts = {}) {
    if (opts && typeof opts === 'object') {
      this.setOptions(opts);
@@ -1093,9 +1225,13 @@ class BaseClient {
        return message;
      }

-      const files = await getFiles({
-        file_id: { $in: fileIds },
-      });
+      const files = await getFiles(
+        {
+          file_id: { $in: fileIds },
+        },
+        {},
+        {},
+      );

      await this.addImageURLs(message, files, this.visionMode);

--- a/api/app/clients/ChatGPTClient.js
+++ b/api/app/clients/ChatGPTClient.js
@@ -1,804 +0,0 @@
-const Keyv = require('keyv');
-const crypto = require('crypto');
-const { CohereClient } = require('cohere-ai');
-const { fetchEventSource } = require('@waylaidwanderer/fetch-event-source');
-const { encoding_for_model: encodingForModel, get_encoding: getEncoding } = require('tiktoken');
-const {
-  ImageDetail,
-  EModelEndpoint,
-  resolveHeaders,
-  CohereConstants,
-  mapModelToAzureConfig,
-} = require('librechat-data-provider');
-const { extractBaseURL, constructAzureURL, genAzureChatCompletion } = require('~/utils');
-const { createContextHandlers } = require('./prompts');
-const { createCoherePayload } = require('./llm');
-const BaseClient = require('./BaseClient');
-const { logger } = require('~/config');
-
-const CHATGPT_MODEL = 'gpt-3.5-turbo';
-const tokenizersCache = {};
-
-class ChatGPTClient extends BaseClient {
-  constructor(apiKey, options = {}, cacheOptions = {}) {
-    super(apiKey, options, cacheOptions);
-
-    cacheOptions.namespace = cacheOptions.namespace || 'chatgpt';
-    this.conversationsCache = new Keyv(cacheOptions);
-    this.setOptions(options);
-  }
-
-  setOptions(options) {
-    if (this.options && !this.options.replaceOptions) {
-      // nested options aren't spread properly, so we need to do this manually
-      this.options.modelOptions = {
-        ...this.options.modelOptions,
-        ...options.modelOptions,
-      };
-      delete options.modelOptions;
-      // now we can merge options
-      this.options = {
-        ...this.options,
-        ...options,
-      };
-    } else {
-      this.options = options;
-    }
-
-    if (this.options.openaiApiKey) {
-      this.apiKey = this.options.openaiApiKey;
-    }
-
-    const modelOptions = this.options.modelOptions || {};
-    this.modelOptions = {
-      ...modelOptions,
-      // set some good defaults (check for undefined in some cases because they may be 0)
-      model: modelOptions.model || CHATGPT_MODEL,
-      temperature: typeof modelOptions.temperature === 'undefined' ? 0.8 : modelOptions.temperature,
-      top_p: typeof modelOptions.top_p === 'undefined' ? 1 : modelOptions.top_p,
-      presence_penalty:
-        typeof modelOptions.presence_penalty === 'undefined' ? 1 : modelOptions.presence_penalty,
-      stop: modelOptions.stop,
-    };
-
-    this.isChatGptModel = this.modelOptions.model.includes('gpt-');
-    const { isChatGptModel } = this;
-    this.isUnofficialChatGptModel =
-      this.modelOptions.model.startsWith('text-chat') ||
-      this.modelOptions.model.startsWith('text-davinci-002-render');
-    const { isUnofficialChatGptModel } = this;
-
-    // Davinci models have a max context length of 4097 tokens.
-    this.maxContextTokens = this.options.maxContextTokens || (isChatGptModel ? 4095 : 4097);
-    // I decided to reserve 1024 tokens for the response.
-    // The max prompt tokens is determined by the max context tokens minus the max response tokens.
-    // Earlier messages will be dropped until the prompt is within the limit.
-    this.maxResponseTokens = this.modelOptions.max_tokens || 1024;
-    this.maxPromptTokens =
-      this.options.maxPromptTokens || this.maxContextTokens - this.maxResponseTokens;
-
-    if (this.maxPromptTokens + this.maxResponseTokens > this.maxContextTokens) {
-      throw new Error(
-        `maxPromptTokens + max_tokens (${this.maxPromptTokens} + ${this.maxResponseTokens} = ${
-          this.maxPromptTokens + this.maxResponseTokens
-        }) must be less than or equal to maxContextTokens (${this.maxContextTokens})`,
-      );
-    }
-
-    this.userLabel = this.options.userLabel || 'User';
-    this.chatGptLabel = this.options.chatGptLabel || 'ChatGPT';
-
-    if (isChatGptModel) {
-      // Use these faux tokens to help the AI understand the context since we are building the chat log ourselves.
-      // Trying to use "<|im_start|>" causes the AI to still generate "<" or "<|" at the end sometimes for some reason,
-      // without tripping the stop sequences, so I'm using "||>" instead.
-      this.startToken = '||>';
-      this.endToken = '';
-      this.gptEncoder = this.constructor.getTokenizer('cl100k_base');
-    } else if (isUnofficialChatGptModel) {
-      this.startToken = '<|im_start|>';
-      this.endToken = '<|im_end|>';
-      this.gptEncoder = this.constructor.getTokenizer('text-davinci-003', true, {
-        '<|im_start|>': 100264,
-        '<|im_end|>': 100265,
-      });
-    } else {
-      // Previously I was trying to use "<|endoftext|>" but there seems to be some bug with OpenAI's token counting
-      // system that causes only the first "<|endoftext|>" to be counted as 1 token, and the rest are not treated
-      // as a single token. So we're using this instead.
-      this.startToken = '||>';
-      this.endToken = '';
-      try {
-        this.gptEncoder = this.constructor.getTokenizer(this.modelOptions.model, true);
-      } catch {
-        this.gptEncoder = this.constructor.getTokenizer('text-davinci-003', true);
-      }
-    }
-
-    if (!this.modelOptions.stop) {
-      const stopTokens = [this.startToken];
-      if (this.endToken && this.endToken !== this.startToken) {
-        stopTokens.push(this.endToken);
-      }
-      stopTokens.push(`\n${this.userLabel}:`);
-      stopTokens.push('<|diff_marker|>');
-      // I chose not to do one for `chatGptLabel` because I've never seen it happen
-      this.modelOptions.stop = stopTokens;
-    }
-
-    if (this.options.reverseProxyUrl) {
-      this.completionsUrl = this.options.reverseProxyUrl;
-    } else if (isChatGptModel) {
-      this.completionsUrl = 'https://api.openai.com/v1/chat/completions';
-    } else {
-      this.completionsUrl = 'https://api.openai.com/v1/completions';
-    }
-
-    return this;
-  }
-
-  static getTokenizer(encoding, isModelName = false, extendSpecialTokens = {}) {
-    if (tokenizersCache[encoding]) {
-      return tokenizersCache[encoding];
-    }
-    let tokenizer;
-    if (isModelName) {
-      tokenizer = encodingForModel(encoding, extendSpecialTokens);
-    } else {
-      tokenizer = getEncoding(encoding, extendSpecialTokens);
-    }
-    tokenizersCache[encoding] = tokenizer;
-    return tokenizer;
-  }
-
-  /** @type {getCompletion} */
-  async getCompletion(input, onProgress, onTokenProgress, abortController = null) {
-    if (!abortController) {
-      abortController = new AbortController();
-    }
-
-    let modelOptions = { ...this.modelOptions };
-    if (typeof onProgress === 'function') {
-      modelOptions.stream = true;
-    }
-    if (this.isChatGptModel) {
-      modelOptions.messages = input;
-    } else {
-      modelOptions.prompt = input;
-    }
-
-    if (this.useOpenRouter && modelOptions.prompt) {
-      delete modelOptions.stop;
-    }
-
-    const { debug } = this.options;
-    let baseURL = this.completionsUrl;
-    if (debug) {
-      console.debug();
-      console.debug(baseURL);
-      console.debug(modelOptions);
-      console.debug();
-    }
-
-    const opts = {
-      method: 'POST',
-      headers: {
-        'Content-Type': 'application/json',
-      },
-    };
-
-    if (this.isVisionModel) {
-      modelOptions.max_tokens = 4000;
-    }
-
-    /** @type {TAzureConfig | undefined} */
-    const azureConfig = this.options?.req?.app?.locals?.[EModelEndpoint.azureOpenAI];
-
-    const isAzure = this.azure || this.options.azure;
-    if (
-      (isAzure && this.isVisionModel && azureConfig) ||
-      (azureConfig && this.isVisionModel && this.options.endpoint === EModelEndpoint.azureOpenAI)
-    ) {
-      const { modelGroupMap, groupMap } = azureConfig;
-      const {
-        azureOptions,
-        baseURL,
-        headers = {},
-        serverless,
-      } = mapModelToAzureConfig({
-        modelName: modelOptions.model,
-        modelGroupMap,
-        groupMap,
-      });
-      opts.headers = resolveHeaders(headers);
-      this.langchainProxy = extractBaseURL(baseURL);
-      this.apiKey = azureOptions.azureOpenAIApiKey;
-
-      const groupName = modelGroupMap[modelOptions.model].group;
-      this.options.addParams = azureConfig.groupMap[groupName].addParams;
-      this.options.dropParams = azureConfig.groupMap[groupName].dropParams;
-      // Note: `forcePrompt` not re-assigned as only chat models are vision models
-
-      this.azure = !serverless && azureOptions;
-      this.azureEndpoint =
-        !serverless && genAzureChatCompletion(this.azure, modelOptions.model, this);
-      if (serverless === true) {
-        this.options.defaultQuery = azureOptions.azureOpenAIApiVersion
-          ? { 'api-version': azureOptions.azureOpenAIApiVersion }
-          : undefined;
-        this.options.headers['api-key'] = this.apiKey;
-      }
-    }
-
-    if (this.options.defaultQuery) {
-      opts.defaultQuery = this.options.defaultQuery;
-    }
-
-    if (this.options.headers) {
-      opts.headers = { ...opts.headers, ...this.options.headers };
-    }
-
-    if (isAzure) {
-      // Azure does not accept `model` in the body, so we need to remove it.
-      delete modelOptions.model;
-
-      baseURL = this.langchainProxy
-        ? constructAzureURL({
-          baseURL: this.langchainProxy,
-          azureOptions: this.azure,
-        })
-        : this.azureEndpoint.split(/(?<!\/)\/(chat|completion)\//)[0];
-
-      if (this.options.forcePrompt) {
-        baseURL += '/completions';
-      } else {
-        baseURL += '/chat/completions';
-      }
-
-      opts.defaultQuery = { 'api-version': this.azure.azureOpenAIApiVersion };
-      opts.headers = { ...opts.headers, 'api-key': this.apiKey };
-    } else if (this.apiKey) {
-      opts.headers.Authorization = `Bearer ${this.apiKey}`;
-    }
-
-    if (process.env.OPENAI_ORGANIZATION) {
-      opts.headers['OpenAI-Organization'] = process.env.OPENAI_ORGANIZATION;
-    }
-
-    if (this.useOpenRouter) {
-      opts.headers['HTTP-Referer'] = 'https://librechat.ai';
-      opts.headers['X-Title'] = 'LibreChat';
-    }
-
-    /* hacky fixes for Mistral AI API:
-      - Re-orders system message to the top of the messages payload, as not allowed anywhere else
-      - If there is only one message and it's a system message, change the role to user
-      */
-    if (baseURL.includes('https://api.mistral.ai/v1') && modelOptions.messages) {
-      const { messages } = modelOptions;
-
-      const systemMessageIndex = messages.findIndex((msg) => msg.role === 'system');
-
-      if (systemMessageIndex > 0) {
-        const [systemMessage] = messages.splice(systemMessageIndex, 1);
-        messages.unshift(systemMessage);
-      }
-
-      modelOptions.messages = messages;
-
-      if (messages.length === 1 && messages[0].role === 'system') {
-        modelOptions.messages[0].role = 'user';
-      }
-    }
-
-    if (this.options.addParams && typeof this.options.addParams === 'object') {
-      modelOptions = {
-        ...modelOptions,
-        ...this.options.addParams,
-      };
-      logger.debug('[ChatGPTClient] chatCompletion: added params', {
-        addParams: this.options.addParams,
-        modelOptions,
-      });
-    }
-
-    if (this.options.dropParams && Array.isArray(this.options.dropParams)) {
-      this.options.dropParams.forEach((param) => {
-        delete modelOptions[param];
-      });
-      logger.debug('[ChatGPTClient] chatCompletion: dropped params', {
-        dropParams: this.options.dropParams,
-        modelOptions,
-      });
-    }
-
-    if (baseURL.startsWith(CohereConstants.API_URL)) {
-      const payload = createCoherePayload({ modelOptions });
-      return await this.cohereChatCompletion({ payload, onTokenProgress });
-    }
-
-    if (baseURL.includes('v1') && !baseURL.includes('/completions') && !this.isChatCompletion) {
-      baseURL = baseURL.split('v1')[0] + 'v1/completions';
-    } else if (
-      baseURL.includes('v1') &&
-      !baseURL.includes('/chat/completions') &&
-      this.isChatCompletion
-    ) {
-      baseURL = baseURL.split('v1')[0] + 'v1/chat/completions';
-    }
-
-    const BASE_URL = new URL(baseURL);
-    if (opts.defaultQuery) {
-      Object.entries(opts.defaultQuery).forEach(([key, value]) => {
-        BASE_URL.searchParams.append(key, value);
-      });
-      delete opts.defaultQuery;
-    }
-
-    const completionsURL = BASE_URL.toString();
-    opts.body = JSON.stringify(modelOptions);
-
-    if (modelOptions.stream) {
-      // eslint-disable-next-line no-async-promise-executor
-      return new Promise(async (resolve, reject) => {
-        try {
-          let done = false;
-          await fetchEventSource(completionsURL, {
-            ...opts,
-            signal: abortController.signal,
-            async onopen(response) {
-              if (response.status === 200) {
-                return;
-              }
-              if (debug) {
-                console.debug(response);
-              }
-              let error;
-              try {
-                const body = await response.text();
-                error = new Error(`Failed to send message. HTTP ${response.status} - ${body}`);
-                error.status = response.status;
-                error.json = JSON.parse(body);
-              } catch {
-                error = error || new Error(`Failed to send message. HTTP ${response.status}`);
-              }
-              throw error;
-            },
-            onclose() {
-              if (debug) {
-                console.debug('Server closed the connection unexpectedly, returning...');
-              }
-              // workaround for private API not sending [DONE] event
-              if (!done) {
-                onProgress('[DONE]');
-                resolve();
-              }
-            },
-            onerror(err) {
-              if (debug) {
-                console.debug(err);
-              }
-              // rethrow to stop the operation
-              throw err;
-            },
-            onmessage(message) {
-              if (debug) {
-                console.debug(message);
-              }
-              if (!message.data || message.event === 'ping') {
-                return;
-              }
-              if (message.data === '[DONE]') {
-                onProgress('[DONE]');
-                resolve();
-                done = true;
-                return;
-              }
-              onProgress(JSON.parse(message.data));
-            },
-          });
-        } catch (err) {
-          reject(err);
-        }
-      });
-    }
-    const response = await fetch(completionsURL, {
-      ...opts,
-      signal: abortController.signal,
-    });
-    if (response.status !== 200) {
-      const body = await response.text();
-      const error = new Error(`Failed to send message. HTTP ${response.status} - ${body}`);
-      error.status = response.status;
-      try {
-        error.json = JSON.parse(body);
-      } catch {
-        error.body = body;
-      }
-      throw error;
-    }
-    return response.json();
-  }
-
-  /** @type {cohereChatCompletion} */
-  async cohereChatCompletion({ payload, onTokenProgress }) {
-    const cohere = new CohereClient({
-      token: this.apiKey,
-      environment: this.completionsUrl,
-    });
-
-    if (!payload.stream) {
-      const chatResponse = await cohere.chat(payload);
-      return chatResponse.text;
-    }
-
-    const chatStream = await cohere.chatStream(payload);
-    let reply = '';
-    for await (const message of chatStream) {
-      if (!message) {
-        continue;
-      }
-
-      if (message.eventType === 'text-generation' && message.text) {
-        onTokenProgress(message.text);
-        reply += message.text;
-      }
-      /*
-      Cohere API Chinese Unicode character replacement hotfix.
-      Should be un-commented when the following issue is resolved:
-      https://github.com/cohere-ai/cohere-typescript/issues/151
-
-      else if (message.eventType === 'stream-end' && message.response) {
-        reply = message.response.text;
-      }
-      */
-    }
-
-    return reply;
-  }
-
-  async generateTitle(userMessage, botMessage) {
-    const instructionsPayload = {
-      role: 'system',
-      content: `Write an extremely concise subtitle for this conversation with no more than a few words. All words should be capitalized. Exclude punctuation.
-
-||>Message:
-${userMessage.message}
-||>Response:
-${botMessage.message}
-
-||>Title:`,
-    };
-
-    const titleGenClientOptions = JSON.parse(JSON.stringify(this.options));
-    titleGenClientOptions.modelOptions = {
-      model: 'gpt-3.5-turbo',
-      temperature: 0,
-      presence_penalty: 0,
-      frequency_penalty: 0,
-    };
-    const titleGenClient = new ChatGPTClient(this.apiKey, titleGenClientOptions);
-    const result = await titleGenClient.getCompletion([instructionsPayload], null);
-    // remove any non-alphanumeric characters, replace multiple spaces with 1, and then trim
-    return result.choices[0].message.content
-      .replace(/[^a-zA-Z0-9' ]/g, '')
-      .replace(/\s+/g, ' ')
-      .trim();
-  }
-
-  async sendMessage(message, opts = {}) {
-    if (opts.clientOptions && typeof opts.clientOptions === 'object') {
-      this.setOptions(opts.clientOptions);
-    }
-
-    const conversationId = opts.conversationId || crypto.randomUUID();
-    const parentMessageId = opts.parentMessageId || crypto.randomUUID();
-
-    let conversation =
-      typeof opts.conversation === 'object'
-        ? opts.conversation
-        : await this.conversationsCache.get(conversationId);
-
-    let isNewConversation = false;
-    if (!conversation) {
-      conversation = {
-        messages: [],
-        createdAt: Date.now(),
-      };
-      isNewConversation = true;
-    }
-
-    const shouldGenerateTitle = opts.shouldGenerateTitle && isNewConversation;
-
-    const userMessage = {
-      id: crypto.randomUUID(),
-      parentMessageId,
-      role: 'User',
-      message,
-    };
-    conversation.messages.push(userMessage);
-
-    // Doing it this way instead of having each message be a separate element in the array seems to be more reliable,
-    // especially when it comes to keeping the AI in character. It also seems to improve coherency and context retention.
-    const { prompt: payload, context } = await this.buildPrompt(
-      conversation.messages,
-      userMessage.id,
-      {
-        isChatGptModel: this.isChatGptModel,
-        promptPrefix: opts.promptPrefix,
-      },
-    );
-
-    if (this.options.keepNecessaryMessagesOnly) {
-      conversation.messages = context;
-    }
-
-    let reply = '';
-    let result = null;
-    if (typeof opts.onProgress === 'function') {
-      await this.getCompletion(
-        payload,
-        (progressMessage) => {
-          if (progressMessage === '[DONE]') {
-            return;
-          }
-          const token = this.isChatGptModel
-            ? progressMessage.choices[0].delta.content
-            : progressMessage.choices[0].text;
-          // first event's delta content is always undefined
-          if (!token) {
-            return;
-          }
-          if (this.options.debug) {
-            console.debug(token);
-          }
-          if (token === this.endToken) {
-            return;
-          }
-          opts.onProgress(token);
-          reply += token;
-        },
-        opts.abortController || new AbortController(),
-      );
-    } else {
-      result = await this.getCompletion(
-        payload,
-        null,
-        opts.abortController || new AbortController(),
-      );
-      if (this.options.debug) {
-        console.debug(JSON.stringify(result));
-      }
-      if (this.isChatGptModel) {
-        reply = result.choices[0].message.content;
-      } else {
-        reply = result.choices[0].text.replace(this.endToken, '');
-      }
-    }
-
-    // avoids some rendering issues when using the CLI app
-    if (this.options.debug) {
-      console.debug();
-    }
-
-    reply = reply.trim();
-
-    const replyMessage = {
-      id: crypto.randomUUID(),
-      parentMessageId: userMessage.id,
-      role: 'ChatGPT',
-      message: reply,
-    };
-    conversation.messages.push(replyMessage);
-
-    const returnData = {
-      response: replyMessage.message,
-      conversationId,
-      parentMessageId: replyMessage.parentMessageId,
-      messageId: replyMessage.id,
-      details: result || {},
-    };
-
-    if (shouldGenerateTitle) {
-      conversation.title = await this.generateTitle(userMessage, replyMessage);
-      returnData.title = conversation.title;
-    }
-
-    await this.conversationsCache.set(conversationId, conversation);
-
-    if (this.options.returnConversation) {
-      returnData.conversation = conversation;
-    }
-
-    return returnData;
-  }
-
-  async buildPrompt(messages, { isChatGptModel = false, promptPrefix = null }) {
-    promptPrefix = (promptPrefix || this.options.promptPrefix || '').trim();
-
-    // Handle attachments and create augmentedPrompt
-    if (this.options.attachments) {
-      const attachments = await this.options.attachments;
-      const lastMessage = messages[messages.length - 1];
-
-      if (this.message_file_map) {
-        this.message_file_map[lastMessage.messageId] = attachments;
-      } else {
-        this.message_file_map = {
-          [lastMessage.messageId]: attachments,
-        };
-      }
-
-      const files = await this.addImageURLs(lastMessage, attachments);
-      this.options.attachments = files;
-
-      this.contextHandlers = createContextHandlers(this.options.req, lastMessage.text);
-    }
-
-    if (this.message_file_map) {
-      this.contextHandlers = createContextHandlers(
-        this.options.req,
-        messages[messages.length - 1].text,
-      );
-    }
-
-    // Calculate image token cost and process embedded files
-    messages.forEach((message, i) => {
-      if (this.message_file_map && this.message_file_map[message.messageId]) {
-        const attachments = this.message_file_map[message.messageId];
-        for (const file of attachments) {
-          if (file.embedded) {
-            this.contextHandlers?.processFile(file);
-            continue;
-          }
-
-          messages[i].tokenCount =
-            (messages[i].tokenCount || 0) +
-            this.calculateImageTokenCost({
-              width: file.width,
-              height: file.height,
-              detail: this.options.imageDetail ?? ImageDetail.auto,
-            });
-        }
-      }
-    });
-
-    if (this.contextHandlers) {
-      this.augmentedPrompt = await this.contextHandlers.createContext();
-      promptPrefix = this.augmentedPrompt + promptPrefix;
-    }
-
-    if (promptPrefix) {
-      // If the prompt prefix doesn't end with the end token, add it.
-      if (!promptPrefix.endsWith(`${this.endToken}`)) {
-        promptPrefix = `${promptPrefix.trim()}${this.endToken}\n\n`;
-      }
-      promptPrefix = `${this.startToken}Instructions:\n${promptPrefix}`;
-    }
-    const promptSuffix = `${this.startToken}${this.chatGptLabel}:\n`; // Prompt ChatGPT to respond.
-
-    const instructionsPayload = {
-      role: 'system',
-      content: promptPrefix,
-    };
-
-    const messagePayload = {
-      role: 'system',
-      content: promptSuffix,
-    };
-
-    let currentTokenCount;
-    if (isChatGptModel) {
-      currentTokenCount =
-        this.getTokenCountForMessage(instructionsPayload) +
-        this.getTokenCountForMessage(messagePayload);
-    } else {
-      currentTokenCount = this.getTokenCount(`${promptPrefix}${promptSuffix}`);
-    }
-    let promptBody = '';
-    const maxTokenCount = this.maxPromptTokens;
-
-    const context = [];
-
-    // Iterate backwards through the messages, adding them to the prompt until we reach the max token count.
-    // Do this within a recursive async function so that it doesn't block the event loop for too long.
-    const buildPromptBody = async () => {
-      if (currentTokenCount < maxTokenCount && messages.length > 0) {
-        const message = messages.pop();
-        const roleLabel =
-          message?.isCreatedByUser || message?.role?.toLowerCase() === 'user'
-            ? this.userLabel
-            : this.chatGptLabel;
-        const messageString = `${this.startToken}${roleLabel}:\n${
-          message?.text ?? message?.message
-        }${this.endToken}\n`;
-        let newPromptBody;
-        if (promptBody || isChatGptModel) {
-          newPromptBody = `${messageString}${promptBody}`;
-        } else {
-          // Always insert prompt prefix before the last user message, if not gpt-3.5-turbo.
-          // This makes the AI obey the prompt instructions better, which is important for custom instructions.
-          // After a bunch of testing, it doesn't seem to cause the AI any confusion, even if you ask it things
-          // like "what's the last thing I wrote?".
-          newPromptBody = `${promptPrefix}${messageString}${promptBody}`;
-        }
-
-        context.unshift(message);
-
-        const tokenCountForMessage = this.getTokenCount(messageString);
-        const newTokenCount = currentTokenCount + tokenCountForMessage;
-        if (newTokenCount > maxTokenCount) {
-          if (promptBody) {
-            // This message would put us over the token limit, so don't add it.
-            return false;
-          }
-          // This is the first message, so we can't add it. Just throw an error.
-          throw new Error(
-            `Prompt is too long. Max token count is ${maxTokenCount}, but prompt is ${newTokenCount} tokens long.`,
-          );
-        }
-        promptBody = newPromptBody;
-        currentTokenCount = newTokenCount;
-        // wait for next tick to avoid blocking the event loop
-        await new Promise((resolve) => setImmediate(resolve));
-        return buildPromptBody();
-      }
-      return true;
-    };
-
-    await buildPromptBody();
-
-    const prompt = `${promptBody}${promptSuffix}`;
-    if (isChatGptModel) {
-      messagePayload.content = prompt;
-      // Add 3 tokens for Assistant Label priming after all messages have been counted.
-      currentTokenCount += 3;
-    }
-
-    // Use up to `this.maxContextTokens` tokens (prompt + response), but try to leave `this.maxTokens` tokens for the response.
-    this.modelOptions.max_tokens = Math.min(
-      this.maxContextTokens - currentTokenCount,
-      this.maxResponseTokens,
-    );
-
-    if (isChatGptModel) {
-      return { prompt: [instructionsPayload, messagePayload], context };
-    }
-    return { prompt, context, promptTokens: currentTokenCount };
-  }
-
-  getTokenCount(text) {
-    return this.gptEncoder.encode(text, 'all').length;
-  }
-
-  /**
-   * Algorithm adapted from "6. Counting tokens for chat API calls" of
-   * https://github.com/openai/openai-cookbook/blob/main/examples/How_to_count_tokens_with_tiktoken.ipynb
-   *
-   * An additional 3 tokens need to be added for assistant label priming after all messages have been counted.
-   *
-   * @param {Object} message
-   */
-  getTokenCountForMessage(message) {
-    // Note: gpt-3.5-turbo and gpt-4 may update over time. Use default for these as well as for unknown models
-    let tokensPerMessage = 3;
-    let tokensPerName = 1;
-
-    if (this.modelOptions.model === 'gpt-3.5-turbo-0301') {
-      tokensPerMessage = 4;
-      tokensPerName = -1;
-    }
-
-    let numTokens = tokensPerMessage;
-    for (let [key, value] of Object.entries(message)) {
-      numTokens += this.getTokenCount(value);
-      if (key === 'name') {
-        numTokens += tokensPerName;
-      }
-    }
-
-    return numTokens;
-  }
-}
-
-module.exports = ChatGPTClient;
--- a/api/app/clients/GoogleClient.js
+++ b/api/app/clients/GoogleClient.js
@@ -1,6 +1,7 @@
 const { google } = require('googleapis');
 const { concat } = require('@langchain/core/utils/stream');
 const { ChatVertexAI } = require('@langchain/google-vertexai');
+const { Tokenizer, getSafetySettings } = require('@librechat/api');
 const { ChatGoogleGenerativeAI } = require('@langchain/google-genai');
 const { GoogleGenerativeAI: GenAI } = require('@google/generative-ai');
 const { HumanMessage, SystemMessage } = require('@langchain/core/messages');
@@ -9,16 +10,16 @@ const {
  validateVisionModel,
  getResponseSender,
  endpointSettings,
+  parseTextParts,
  EModelEndpoint,
+  googleSettings,
  ContentTypes,
  VisionModes,
  ErrorTypes,
  Constants,
  AuthKeys,
 } = require('librechat-data-provider');
-const { getSafetySettings } = require('~/server/services/Endpoints/google/llm');
 const { encodeAndFormat } = require('~/server/services/Files/images');
-const Tokenizer = require('~/server/services/Tokenizer');
 const { spendTokens } = require('~/models/spendTokens');
 const { getModelMaxTokens } = require('~/utils');
 const { sleep } = require('~/server/utils');
@@ -33,7 +34,8 @@ const BaseClient = require('./BaseClient');

 const loc = process.env.GOOGLE_LOC || 'us-central1';
 const publisher = 'google';
-const endpointPrefix = `${loc}-aiplatform.googleapis.com`;
+const endpointPrefix =
+  loc === 'global' ? 'aiplatform.googleapis.com' : `${loc}-aiplatform.googleapis.com`;

 const settings = endpointSettings[EModelEndpoint.google];
 const EXCLUDED_GENAI_MODELS = /gemini-(?:1\.0|1-0|pro)/;
@@ -139,8 +141,7 @@ class GoogleClient extends BaseClient {
    this.options.attachments?.then((attachments) => this.checkVisionRequest(attachments));

    /** @type {boolean} Whether using a "GenerativeAI" Model */
-    this.isGenerativeModel =
-      this.modelOptions.model.includes('gemini') || this.modelOptions.model.includes('learnlm');
+    this.isGenerativeModel = /gemini|learnlm|gemma/.test(this.modelOptions.model);

    this.maxContextTokens =
      this.options.maxContextTokens ??
@@ -165,6 +166,16 @@ class GoogleClient extends BaseClient {
      );
    }

+    // Add thinking configuration
+    this.modelOptions.thinkingConfig = {
+      thinkingBudget:
+        (this.modelOptions.thinking ?? googleSettings.thinking.default)
+          ? this.modelOptions.thinkingBudget
+          : 0,
+    };
+    delete this.modelOptions.thinking;
+    delete this.modelOptions.thinkingBudget;
+
    this.sender =
      this.options.sender ??
      getResponseSender({
@@ -198,7 +209,11 @@ class GoogleClient extends BaseClient {
   */
  checkVisionRequest(attachments) {
    /* Validation vision request */
-    this.defaultVisionModel = this.options.visionModel ?? 'gemini-pro-vision';
+    this.defaultVisionModel =
+      this.options.visionModel ??
+      (!EXCLUDED_GENAI_MODELS.test(this.modelOptions.model)
+        ? this.modelOptions.model
+        : 'gemini-pro-vision');
    const availableModels = this.options.modelsConfig?.[EModelEndpoint.google];
    this.isVisionModel = validateVisionModel({ model: this.modelOptions.model, availableModels });

@@ -232,11 +247,11 @@ class GoogleClient extends BaseClient {
      msg.content = (
        !Array.isArray(msg.content)
          ? [
-            {
-              type: ContentTypes.TEXT,
-              [ContentTypes.TEXT]: msg.content,
-            },
-          ]
+              {
+                type: ContentTypes.TEXT,
+                [ContentTypes.TEXT]: msg.content,
+              },
+            ]
          : msg.content
      ).concat(message.image_urls);

@@ -313,6 +328,9 @@ class GoogleClient extends BaseClient {
          this.contextHandlers?.processFile(file);
          continue;
        }
+        if (file.metadata?.fileIdentifier) {
+          continue;
+        }
      }

      this.augmentedPrompt = await this.contextHandlers.createContext();
@@ -770,6 +788,22 @@ class GoogleClient extends BaseClient {
    return this.usage;
  }

+  getMessageMapMethod() {
+    /**
+     * @param {TMessage} msg
+     */
+    return (msg) => {
+      if (msg.text != null && msg.text && msg.text.startsWith(':::thinking')) {
+        msg.text = msg.text.replace(/:::thinking.*?:::/gs, '').trim();
+      } else if (msg.content != null) {
+        msg.text = parseTextParts(msg.content, true);
+        delete msg.content;
+      }
+
+      return msg;
+    };
+  }
+
  /**
   * Calculates the correct token count for the current user message based on the token count map and API usage.
   * Edge case: If the calculation results in a negative value, it returns the original estimate.
@@ -827,7 +861,8 @@ class GoogleClient extends BaseClient {
    let reply = '';
    const { abortController } = options;

-    const model = this.modelOptions.modelName ?? this.modelOptions.model ?? '';
+    const model =
+      this.options.titleModel ?? this.modelOptions.modelName ?? this.modelOptions.model ?? '';
    const safetySettings = getSafetySettings(model);
    if (!EXCLUDED_GENAI_MODELS.test(model) && !this.project_id) {
      logger.debug('Identified titling model as GenAI version');
--- a/api/app/clients/OllamaClient.js
+++ b/api/app/clients/OllamaClient.js
@@ -1,10 +1,11 @@
 const { z } = require('zod');
 const axios = require('axios');
 const { Ollama } = require('ollama');
+const { sleep } = require('@librechat/agents');
+const { logAxiosError } = require('@librechat/api');
+const { logger } = require('@librechat/data-schemas');
 const { Constants } = require('librechat-data-provider');
-const { deriveBaseURL, logAxiosError } = require('~/utils');
-const { sleep } = require('~/server/utils');
-const { logger } = require('~/config');
+const { deriveBaseURL } = require('~/utils');

 const ollamaPayloadSchema = z.object({
  mirostat: z.number().optional(),
@@ -67,7 +68,7 @@ class OllamaClient {
      return models;
    } catch (error) {
      const logMessage =
-        'Failed to fetch models from Ollama API. If you are not using Ollama directly, and instead, through some aggregator or reverse proxy that handles fetching via OpenAI spec, ensure the name of the endpoint doesn\'t start with `ollama` (case-insensitive).';
+        "Failed to fetch models from Ollama API. If you are not using Ollama directly, and instead, through some aggregator or reverse proxy that handles fetching via OpenAI spec, ensure the name of the endpoint doesn't start with `ollama` (case-insensitive).";
      logAxiosError({ message: logMessage, error });
      return [];
    }
--- a/api/app/clients/OpenAIClient.js
+++ b/api/app/clients/OpenAIClient.js
@@ -1,12 +1,21 @@
-const OpenAI = require('openai');
 const { OllamaClient } = require('./OllamaClient');
 const { HttpsProxyAgent } = require('https-proxy-agent');
-const { SplitStreamHandler, GraphEvents } = require('@librechat/agents');
+const { SplitStreamHandler, CustomOpenAIClient: OpenAI } = require('@librechat/agents');
+const {
+  isEnabled,
+  Tokenizer,
+  createFetch,
+  resolveHeaders,
+  constructAzureURL,
+  genAzureChatCompletion,
+  createStreamEventHandlers,
+} = require('@librechat/api');
 const {
  Constants,
  ImageDetail,
+  ContentTypes,
+  parseTextParts,
  EModelEndpoint,
-  resolveHeaders,
  KnownEndpoints,
  openAISettings,
  ImageDetailCost,
@@ -15,13 +24,6 @@ const {
  validateVisionModel,
  mapModelToAzureConfig,
 } = require('librechat-data-provider');
-const {
-  extractBaseURL,
-  constructAzureURL,
-  getModelMaxTokens,
-  genAzureChatCompletion,
-  getModelMaxOutputTokens,
-} = require('~/utils');
 const {
  truncateText,
  formatMessage,
@@ -29,28 +31,21 @@ const {
  titleInstruction,
  createContextHandlers,
 } = require('./prompts');
+const { extractBaseURL, getModelMaxTokens, getModelMaxOutputTokens } = require('~/utils');
 const { encodeAndFormat } = require('~/server/services/Files/images/encode');
-const { addSpaceIfNeeded, isEnabled, sleep } = require('~/server/utils');
-const Tokenizer = require('~/server/services/Tokenizer');
+const { addSpaceIfNeeded, sleep } = require('~/server/utils');
 const { spendTokens } = require('~/models/spendTokens');
 const { handleOpenAIErrors } = require('./tools/util');
 const { createLLM, RunManager } = require('./llm');
-const { logger, sendEvent } = require('~/config');
-const ChatGPTClient = require('./ChatGPTClient');
 const { summaryBuffer } = require('./memory');
 const { runTitleChain } = require('./chains');
 const { tokenSplit } = require('./document');
 const BaseClient = require('./BaseClient');
+const { logger } = require('~/config');

 class OpenAIClient extends BaseClient {
  constructor(apiKey, options = {}) {
    super(apiKey, options);
-    this.ChatGPTClient = new ChatGPTClient();
-    this.buildPrompt = this.ChatGPTClient.buildPrompt.bind(this);
-    /** @type {getCompletion} */
-    this.getCompletion = this.ChatGPTClient.getCompletion.bind(this);
-    /** @type {cohereChatCompletion} */
-    this.cohereChatCompletion = this.ChatGPTClient.cohereChatCompletion.bind(this);
    this.contextStrategy = options.contextStrategy
      ? options.contextStrategy.toLowerCase()
      : 'discard';
@@ -106,18 +101,18 @@ class OpenAIClient extends BaseClient {
      this.checkVisionRequest(this.options.attachments);
    }

-    const omniPattern = /\b(o1|o3)\b/i;
+    const omniPattern = /\b(o\d)\b/i;
    this.isOmni = omniPattern.test(this.modelOptions.model);

-    const { OPENROUTER_API_KEY, OPENAI_FORCE_PROMPT } = process.env ?? {};
-    if (OPENROUTER_API_KEY && !this.azure) {
-      this.apiKey = OPENROUTER_API_KEY;
-      this.useOpenRouter = true;
-    }
-
+    const { OPENAI_FORCE_PROMPT } = process.env ?? {};
    const { reverseProxyUrl: reverseProxy } = this.options;

-    if (!this.useOpenRouter && reverseProxy && reverseProxy.includes(KnownEndpoints.openrouter)) {
+    if (
+      !this.useOpenRouter &&
+      ((reverseProxy && reverseProxy.includes(KnownEndpoints.openrouter)) ||
+        (this.options.endpoint &&
+          this.options.endpoint.toLowerCase().includes(KnownEndpoints.openrouter)))
+    ) {
      this.useOpenRouter = true;
    }

@@ -225,10 +220,6 @@ class OpenAIClient extends BaseClient {
      logger.debug('Using Azure endpoint');
    }

-    if (this.useOpenRouter) {
-      this.completionsUrl = 'https://openrouter.ai/api/v1/chat/completions';
-    }
-
    return this;
  }

@@ -303,7 +294,9 @@ class OpenAIClient extends BaseClient {
  }

  getEncoding() {
-    return this.model?.includes('gpt-4o') ? 'o200k_base' : 'cl100k_base';
+    return this.modelOptions?.model && /gpt-4[^-\s]/.test(this.modelOptions.model)
+      ? 'o200k_base'
+      : 'cl100k_base';
  }

  /**
@@ -379,23 +372,12 @@ class OpenAIClient extends BaseClient {
    return files;
  }

-  async buildMessages(
-    messages,
-    parentMessageId,
-    { isChatCompletion = false, promptPrefix = null },
-    opts,
-  ) {
+  async buildMessages(messages, parentMessageId, { promptPrefix = null }, opts) {
    let orderedMessages = this.constructor.getMessagesForConversation({
      messages,
      parentMessageId,
      summary: this.shouldSummarize,
    });
-    if (!isChatCompletion) {
-      return await this.buildPrompt(orderedMessages, {
-        isChatGptModel: isChatCompletion,
-        promptPrefix,
-      });
-    }

    let payload;
    let instructions;
@@ -455,6 +437,9 @@ class OpenAIClient extends BaseClient {
            this.contextHandlers?.processFile(file);
            continue;
          }
+          if (file.metadata?.fileIdentifier) {
+            continue;
+          }

          orderedMessages[i].tokenCount += this.calculateImageTokenCost({
            width: file.width,
@@ -472,7 +457,9 @@ class OpenAIClient extends BaseClient {
      promptPrefix = this.augmentedPrompt + promptPrefix;
    }

-    if (promptPrefix && this.isOmni !== true) {
+    const noSystemModelRegex = /\b(o1-preview|o1-mini)\b/i.test(this.modelOptions.model);
+
+    if (promptPrefix && !noSystemModelRegex) {
      promptPrefix = `Instructions:\n${promptPrefix.trim()}`;
      instructions = {
        role: 'system',
@@ -500,11 +487,27 @@ class OpenAIClient extends BaseClient {
    };

    /** EXPERIMENTAL */
-    if (promptPrefix && this.isOmni === true) {
+    if (promptPrefix && noSystemModelRegex) {
      const lastUserMessageIndex = payload.findLastIndex((message) => message.role === 'user');
      if (lastUserMessageIndex !== -1) {
-        payload[lastUserMessageIndex].content =
-          `${promptPrefix}\n${payload[lastUserMessageIndex].content}`;
+        if (Array.isArray(payload[lastUserMessageIndex].content)) {
+          const firstTextPartIndex = payload[lastUserMessageIndex].content.findIndex(
+            (part) => part.type === ContentTypes.TEXT,
+          );
+          if (firstTextPartIndex !== -1) {
+            const firstTextPart = payload[lastUserMessageIndex].content[firstTextPartIndex];
+            payload[lastUserMessageIndex].content[firstTextPartIndex].text =
+              `${promptPrefix}\n${firstTextPart.text}`;
+          } else {
+            payload[lastUserMessageIndex].content.unshift({
+              type: ContentTypes.TEXT,
+              text: promptPrefix,
+            });
+          }
+        } else {
+          payload[lastUserMessageIndex].content =
+            `${promptPrefix}\n${payload[lastUserMessageIndex].content}`;
+        }
      }
    }

@@ -593,7 +596,7 @@ class OpenAIClient extends BaseClient {
        return result.trim();
      }

-      logger.debug('[OpenAIClient] sendCompletion: result', result);
+      logger.debug('[OpenAIClient] sendCompletion: result', { ...result });

      if (this.isChatCompletion) {
        reply = result.choices[0].message.content;
@@ -610,7 +613,7 @@ class OpenAIClient extends BaseClient {
  }

  initializeLLM({
-    model = 'gpt-4o-mini',
+    model = openAISettings.model.default,
    modelName,
    temperature = 0.2,
    max_tokens,
@@ -711,7 +714,7 @@ class OpenAIClient extends BaseClient {

    const { OPENAI_TITLE_MODEL } = process.env ?? {};

-    let model = this.options.titleModel ?? OPENAI_TITLE_MODEL ?? 'gpt-4o-mini';
+    let model = this.options.titleModel ?? OPENAI_TITLE_MODEL ?? openAISettings.model.default;
    if (model === Constants.CURRENT_MODEL) {
      model = this.modelOptions.model;
    }
@@ -802,7 +805,7 @@ ${convo}

        const completionTokens = this.getTokenCount(title);

-        this.recordTokenUsage({ promptTokens, completionTokens, context: 'title' });
+        await this.recordTokenUsage({ promptTokens, completionTokens, context: 'title' });
      } catch (e) {
        logger.error(
          '[OpenAIClient] There was an issue generating the title with the completion method',
@@ -904,7 +907,7 @@ ${convo}
    let prompt;

    // TODO: remove the gpt fallback and make it specific to endpoint
-    const { OPENAI_SUMMARY_MODEL = 'gpt-4o-mini' } = process.env ?? {};
+    const { OPENAI_SUMMARY_MODEL = openAISettings.model.default } = process.env ?? {};
    let model = this.options.summaryModel ?? OPENAI_SUMMARY_MODEL;
    if (model === Constants.CURRENT_MODEL) {
      model = this.modelOptions.model;
@@ -1105,6 +1108,9 @@ ${convo}
    return (msg) => {
      if (msg.text != null && msg.text && msg.text.startsWith(':::thinking')) {
        msg.text = msg.text.replace(/:::thinking.*?:::/gs, '').trim();
+      } else if (msg.content != null) {
+        msg.text = parseTextParts(msg.content, true);
+        delete msg.content;
      }

      return msg;
@@ -1135,6 +1141,7 @@ ${convo}
      logger.debug('[OpenAIClient] chatCompletion', { baseURL, modelOptions });
      const opts = {
        baseURL,
+        fetchOptions: {},
      };

      if (this.useOpenRouter) {
@@ -1153,11 +1160,7 @@ ${convo}
      }

      if (this.options.proxy) {
-        opts.httpAgent = new HttpsProxyAgent(this.options.proxy);
-      }
-
-      if (this.isVisionModel) {
-        modelOptions.max_tokens = 4000;
+        opts.fetchOptions.agent = new HttpsProxyAgent(this.options.proxy);
      }

      /** @type {TAzureConfig | undefined} */
@@ -1209,9 +1212,9 @@ ${convo}

        opts.baseURL = this.langchainProxy
          ? constructAzureURL({
-            baseURL: this.langchainProxy,
-            azureOptions: this.azure,
-          })
+              baseURL: this.langchainProxy,
+              azureOptions: this.azure,
+            })
          : this.azureEndpoint.split(/(?<!\/)\/(chat|completion)\//)[0];

        opts.defaultQuery = { 'api-version': this.azure.azureOpenAIApiVersion };
@@ -1222,6 +1225,9 @@ ${convo}
        modelOptions.max_completion_tokens = modelOptions.max_tokens;
        delete modelOptions.max_tokens;
      }
+      if (this.isOmni === true && modelOptions.temperature != null) {
+        delete modelOptions.temperature;
+      }

      if (process.env.OPENAI_ORGANIZATION) {
        opts.organization = process.env.OPENAI_ORGANIZATION;
@@ -1230,7 +1236,10 @@ ${convo}
      let chatCompletion;
      /** @type {OpenAI} */
      const openai = new OpenAI({
-        fetch: this.fetch,
+        fetch: createFetch({
+          directEndpoint: this.options.directEndpoint,
+          reverseProxyUrl: this.options.reverseProxyUrl,
+        }),
        apiKey: this.apiKey,
        ...opts,
      });
@@ -1259,23 +1268,56 @@ ${convo}
        modelOptions.messages[0].role = 'user';
      }

+      if (
+        (this.options.endpoint === EModelEndpoint.openAI ||
+          this.options.endpoint === EModelEndpoint.azureOpenAI) &&
+        modelOptions.stream === true
+      ) {
+        modelOptions.stream_options = { include_usage: true };
+      }
+
      if (this.options.addParams && typeof this.options.addParams === 'object') {
+        const addParams = { ...this.options.addParams };
        modelOptions = {
          ...modelOptions,
-          ...this.options.addParams,
+          ...addParams,
        };
        logger.debug('[OpenAIClient] chatCompletion: added params', {
-          addParams: this.options.addParams,
+          addParams: addParams,
          modelOptions,
        });
      }

+      /** Note: OpenAI Web Search models do not support any known parameters besdies `max_tokens` */
+      if (modelOptions.model && /gpt-4o.*search/.test(modelOptions.model)) {
+        const searchExcludeParams = [
+          'frequency_penalty',
+          'presence_penalty',
+          'temperature',
+          'top_p',
+          'top_k',
+          'stop',
+          'logit_bias',
+          'seed',
+          'response_format',
+          'n',
+          'logprobs',
+          'user',
+        ];
+
+        this.options.dropParams = this.options.dropParams || [];
+        this.options.dropParams = [
+          ...new Set([...this.options.dropParams, ...searchExcludeParams]),
+        ];
+      }
+
      if (this.options.dropParams && Array.isArray(this.options.dropParams)) {
-        this.options.dropParams.forEach((param) => {
+        const dropParams = [...this.options.dropParams];
+        dropParams.forEach((param) => {
          delete modelOptions[param];
        });
        logger.debug('[OpenAIClient] chatCompletion: dropped params', {
-          dropParams: this.options.dropParams,
+          dropParams: dropParams,
          modelOptions,
        });
      }
@@ -1298,15 +1340,11 @@ ${convo}
      let streamResolve;

      if (
-        this.isOmni === true &&
-        (this.azure || /o1(?!-(?:mini|preview)).*$/.test(modelOptions.model)) &&
-        !/o3-.*$/.test(this.modelOptions.model) &&
-        modelOptions.stream
+        (!this.isOmni || /^o1-(mini|preview)/i.test(modelOptions.model)) &&
+        modelOptions.reasoning_effort != null
      ) {
-        delete modelOptions.stream;
-        delete modelOptions.stop;
-      } else if (!this.isOmni && modelOptions.reasoning_effort != null) {
        delete modelOptions.reasoning_effort;
+        delete modelOptions.temperature;
      }

      let reasoningKey = 'reasoning_content';
@@ -1314,16 +1352,19 @@ ${convo}
        modelOptions.include_reasoning = true;
        reasoningKey = 'reasoning';
      }
+      if (this.useOpenRouter && modelOptions.reasoning_effort != null) {
+        modelOptions.reasoning = {
+          effort: modelOptions.reasoning_effort,
+        };
+        delete modelOptions.reasoning_effort;
+      }

+      const handlers = createStreamEventHandlers(this.options.res);
      this.streamHandler = new SplitStreamHandler({
        reasoningKey,
        accumulate: true,
        runId: this.responseMessageId,
-        handlers: {
-          [GraphEvents.ON_RUN_STEP]: (event) => sendEvent(this.options.res, event),
-          [GraphEvents.ON_MESSAGE_DELTA]: (event) => sendEvent(this.options.res, event),
-          [GraphEvents.ON_REASONING_DELTA]: (event) => sendEvent(this.options.res, event),
-        },
+        handlers,
      });

      intermediateReply = this.streamHandler.tokens;
@@ -1337,13 +1378,7 @@ ${convo}
          ...modelOptions,
          stream: true,
        };
-        if (
-          this.options.endpoint === EModelEndpoint.openAI ||
-          this.options.endpoint === EModelEndpoint.azureOpenAI
-        ) {
-          params.stream_options = { include_usage: true };
-        }
-        const stream = await openai.beta.chat.completions
+        const stream = await openai.chat.completions
          .stream(params)
          .on('abort', () => {
            /* Do nothing here */
@@ -1427,6 +1462,11 @@ ${convo}
          });
      }

+      if (openai.abortHandler && abortController.signal) {
+        abortController.signal.removeEventListener('abort', openai.abortHandler);
+        openai.abortHandler = undefined;
+      }
+
      if (!chatCompletion && UnexpectedRoleError) {
        throw new Error(
          'OpenAI error: Invalid final message: OpenAI expects final message to include role=assistant',
--- a/api/app/clients/PluginsClient.js
+++ b/api/app/clients/PluginsClient.js
@@ -1,540 +0,0 @@
-const OpenAIClient = require('./OpenAIClient');
-const { CallbackManager } = require('@langchain/core/callbacks/manager');
-const { BufferMemory, ChatMessageHistory } = require('langchain/memory');
-const { addImages, buildErrorInput, buildPromptPrefix } = require('./output_parsers');
-const { initializeCustomAgent, initializeFunctionsAgent } = require('./agents');
-const { processFileURL } = require('~/server/services/Files/process');
-const { EModelEndpoint } = require('librechat-data-provider');
-const { formatLangChainMessages } = require('./prompts');
-const checkBalance = require('~/models/checkBalance');
-const { isEnabled } = require('~/server/utils');
-const { extractBaseURL } = require('~/utils');
-const { loadTools } = require('./tools/util');
-const { logger } = require('~/config');
-
-class PluginsClient extends OpenAIClient {
-  constructor(apiKey, options = {}) {
-    super(apiKey, options);
-    this.sender = options.sender ?? 'Assistant';
-    this.tools = [];
-    this.actions = [];
-    this.setOptions(options);
-    this.openAIApiKey = this.apiKey;
-    this.executor = null;
-  }
-
-  setOptions(options) {
-    this.agentOptions = { ...options.agentOptions };
-    this.functionsAgent = this.agentOptions?.agent === 'functions';
-    this.agentIsGpt3 = this.agentOptions?.model?.includes('gpt-3');
-
-    super.setOptions(options);
-
-    this.isGpt3 = this.modelOptions?.model?.includes('gpt-3');
-
-    if (this.options.reverseProxyUrl) {
-      this.langchainProxy = extractBaseURL(this.options.reverseProxyUrl);
-    }
-  }
-
-  getSaveOptions() {
-    return {
-      artifacts: this.options.artifacts,
-      chatGptLabel: this.options.chatGptLabel,
-      modelLabel: this.options.modelLabel,
-      promptPrefix: this.options.promptPrefix,
-      tools: this.options.tools,
-      ...this.modelOptions,
-      agentOptions: this.agentOptions,
-      iconURL: this.options.iconURL,
-      greeting: this.options.greeting,
-      spec: this.options.spec,
-    };
-  }
-
-  saveLatestAction(action) {
-    this.actions.push(action);
-  }
-
-  getFunctionModelName(input) {
-    if (/-(?!0314)\d{4}/.test(input)) {
-      return input;
-    } else if (input.includes('gpt-3.5-turbo')) {
-      return 'gpt-3.5-turbo';
-    } else if (input.includes('gpt-4')) {
-      return 'gpt-4';
-    } else {
-      return 'gpt-3.5-turbo';
-    }
-  }
-
-  getBuildMessagesOptions(opts) {
-    return {
-      isChatCompletion: true,
-      promptPrefix: opts.promptPrefix,
-      abortController: opts.abortController,
-    };
-  }
-
-  async initialize({ user, message, onAgentAction, onChainEnd, signal }) {
-    const modelOptions = {
-      modelName: this.agentOptions.model,
-      temperature: this.agentOptions.temperature,
-    };
-
-    const model = this.initializeLLM({
-      ...modelOptions,
-      context: 'plugins',
-      initialMessageCount: this.currentMessages.length + 1,
-    });
-
-    logger.debug(
-      `[PluginsClient] Agent Model: ${model.modelName} | Temp: ${model.temperature} | Functions: ${this.functionsAgent}`,
-    );
-
-    // Map Messages to Langchain format
-    const pastMessages = formatLangChainMessages(this.currentMessages.slice(0, -1), {
-      userName: this.options?.name,
-    });
-    logger.debug('[PluginsClient] pastMessages: ' + pastMessages.length);
-
-    // TODO: use readOnly memory, TokenBufferMemory? (both unavailable in LangChainJS)
-    const memory = new BufferMemory({
-      llm: model,
-      chatHistory: new ChatMessageHistory(pastMessages),
-    });
-
-    const { loadedTools } = await loadTools({
-      user,
-      model,
-      tools: this.options.tools,
-      functions: this.functionsAgent,
-      options: {
-        memory,
-        signal: this.abortController.signal,
-        openAIApiKey: this.openAIApiKey,
-        conversationId: this.conversationId,
-        fileStrategy: this.options.req.app.locals.fileStrategy,
-        processFileURL,
-        message,
-      },
-      useSpecs: true,
-    });
-
-    if (loadedTools.length === 0) {
-      return;
-    }
-
-    this.tools = loadedTools;
-
-    logger.debug('[PluginsClient] Requested Tools', this.options.tools);
-    logger.debug(
-      '[PluginsClient] Loaded Tools',
-      this.tools.map((tool) => tool.name),
-    );
-
-    const handleAction = (action, runId, callback = null) => {
-      this.saveLatestAction(action);
-
-      logger.debug('[PluginsClient] Latest Agent Action ', this.actions[this.actions.length - 1]);
-
-      if (typeof callback === 'function') {
-        callback(action, runId);
-      }
-    };
-
-    // initialize agent
-    const initializer = this.functionsAgent ? initializeFunctionsAgent : initializeCustomAgent;
-
-    let customInstructions = (this.options.promptPrefix ?? '').trim();
-    if (typeof this.options.artifactsPrompt === 'string' && this.options.artifactsPrompt) {
-      customInstructions = `${customInstructions ?? ''}\n${this.options.artifactsPrompt}`.trim();
-    }
-
-    this.executor = await initializer({
-      model,
-      signal,
-      pastMessages,
-      tools: this.tools,
-      customInstructions,
-      verbose: this.options.debug,
-      returnIntermediateSteps: true,
-      customName: this.options.chatGptLabel,
-      currentDateString: this.currentDateString,
-      callbackManager: CallbackManager.fromHandlers({
-        async handleAgentAction(action, runId) {
-          handleAction(action, runId, onAgentAction);
-        },
-        async handleChainEnd(action) {
-          if (typeof onChainEnd === 'function') {
-            onChainEnd(action);
-          }
-        },
-      }),
-    });
-
-    logger.debug('[PluginsClient] Loaded agent.');
-  }
-
-  async executorCall(message, { signal, stream, onToolStart, onToolEnd }) {
-    let errorMessage = '';
-    const maxAttempts = 1;
-
-    for (let attempts = 1; attempts <= maxAttempts; attempts++) {
-      const errorInput = buildErrorInput({
-        message,
-        errorMessage,
-        actions: this.actions,
-        functionsAgent: this.functionsAgent,
-      });
-      const input = attempts > 1 ? errorInput : message;
-
-      logger.debug(`[PluginsClient] Attempt ${attempts} of ${maxAttempts}`);
-
-      if (errorMessage.length > 0) {
-        logger.debug('[PluginsClient] Caught error, input: ' + JSON.stringify(input));
-      }
-
-      try {
-        this.result = await this.executor.call({ input, signal }, [
-          {
-            async handleToolStart(...args) {
-              await onToolStart(...args);
-            },
-            async handleToolEnd(...args) {
-              await onToolEnd(...args);
-            },
-            async handleLLMEnd(output) {
-              const { generations } = output;
-              const { text } = generations[0][0];
-              if (text && typeof stream === 'function') {
-                await stream(text);
-              }
-            },
-          },
-        ]);
-        break; // Exit the loop if the function call is successful
-      } catch (err) {
-        logger.error('[PluginsClient] executorCall error:', err);
-        if (attempts === maxAttempts) {
-          const { run } = this.runManager.getRunByConversationId(this.conversationId);
-          const defaultOutput = `Encountered an error while attempting to respond: ${err.message}`;
-          this.result.output = run && run.error ? run.error : defaultOutput;
-          this.result.errorMessage = run && run.error ? run.error : err.message;
-          this.result.intermediateSteps = this.actions;
-          break;
-        }
-      }
-    }
-  }
-
-  /**
-   *
-   * @param {TMessage} responseMessage
-   * @param {Partial<TMessage>} saveOptions
-   * @param {string} user
-   * @returns
-   */
-  async handleResponseMessage(responseMessage, saveOptions, user) {
-    const { output, errorMessage, ...result } = this.result;
-    logger.debug('[PluginsClient][handleResponseMessage] Output:', {
-      output,
-      errorMessage,
-      ...result,
-    });
-    const { error } = responseMessage;
-    if (!error) {
-      responseMessage.tokenCount = this.getTokenCountForResponse(responseMessage);
-      responseMessage.completionTokens = this.getTokenCount(responseMessage.text);
-    }
-
-    // Record usage only when completion is skipped as it is already recorded in the agent phase.
-    if (!this.agentOptions.skipCompletion && !error) {
-      await this.recordTokenUsage(responseMessage);
-    }
-
-    this.responsePromise = this.saveMessageToDatabase(responseMessage, saveOptions, user);
-    delete responseMessage.tokenCount;
-    return { ...responseMessage, ...result };
-  }
-
-  async sendMessage(message, opts = {}) {
-    /** @type {{ filteredTools: string[], includedTools: string[] }} */
-    const { filteredTools = [], includedTools = [] } = this.options.req.app.locals;
-
-    if (includedTools.length > 0) {
-      const tools = this.options.tools.filter((plugin) => includedTools.includes(plugin));
-      this.options.tools = tools;
-    } else {
-      const tools = this.options.tools.filter((plugin) => !filteredTools.includes(plugin));
-      this.options.tools = tools;
-    }
-
-    // If a message is edited, no tools can be used.
-    const completionMode = this.options.tools.length === 0 || opts.isEdited;
-    if (completionMode) {
-      this.setOptions(opts);
-      return super.sendMessage(message, opts);
-    }
-
-    logger.debug('[PluginsClient] sendMessage', { userMessageText: message, opts });
-    const {
-      user,
-      conversationId,
-      responseMessageId,
-      saveOptions,
-      userMessage,
-      onAgentAction,
-      onChainEnd,
-      onToolStart,
-      onToolEnd,
-    } = await this.handleStartMethods(message, opts);
-
-    if (opts.progressCallback) {
-      opts.onProgress = opts.progressCallback.call(null, {
-        ...(opts.progressOptions ?? {}),
-        parentMessageId: userMessage.messageId,
-        messageId: responseMessageId,
-      });
-    }
-
-    this.currentMessages.push(userMessage);
-
-    let {
-      prompt: payload,
-      tokenCountMap,
-      promptTokens,
-    } = await this.buildMessages(
-      this.currentMessages,
-      userMessage.messageId,
-      this.getBuildMessagesOptions({
-        promptPrefix: null,
-        abortController: this.abortController,
-      }),
-    );
-
-    if (tokenCountMap) {
-      logger.debug('[PluginsClient] tokenCountMap', { tokenCountMap });
-      if (tokenCountMap[userMessage.messageId]) {
-        userMessage.tokenCount = tokenCountMap[userMessage.messageId];
-        logger.debug('[PluginsClient] userMessage.tokenCount', userMessage.tokenCount);
-      }
-      this.handleTokenCountMap(tokenCountMap);
-    }
-
-    this.result = {};
-    if (payload) {
-      this.currentMessages = payload;
-    }
-
-    if (!this.skipSaveUserMessage) {
-      this.userMessagePromise = this.saveMessageToDatabase(userMessage, saveOptions, user);
-      if (typeof opts?.getReqData === 'function') {
-        opts.getReqData({
-          userMessagePromise: this.userMessagePromise,
-        });
-      }
-    }
-
-    if (isEnabled(process.env.CHECK_BALANCE)) {
-      await checkBalance({
-        req: this.options.req,
-        res: this.options.res,
-        txData: {
-          user: this.user,
-          tokenType: 'prompt',
-          amount: promptTokens,
-          debug: this.options.debug,
-          model: this.modelOptions.model,
-          endpoint: EModelEndpoint.openAI,
-        },
-      });
-    }
-
-    const responseMessage = {
-      endpoint: EModelEndpoint.gptPlugins,
-      iconURL: this.options.iconURL,
-      messageId: responseMessageId,
-      conversationId,
-      parentMessageId: userMessage.messageId,
-      isCreatedByUser: false,
-      model: this.modelOptions.model,
-      sender: this.sender,
-      promptTokens,
-    };
-
-    await this.initialize({
-      user,
-      message,
-      onAgentAction,
-      onChainEnd,
-      signal: this.abortController.signal,
-      onProgress: opts.onProgress,
-    });
-
-    // const stream = async (text) => {
-    //   await this.generateTextStream.call(this, text, opts.onProgress, { delay: 1 });
-    // };
-    await this.executorCall(message, {
-      signal: this.abortController.signal,
-      // stream,
-      onToolStart,
-      onToolEnd,
-    });
-
-    // If message was aborted mid-generation
-    if (this.result?.errorMessage?.length > 0 && this.result?.errorMessage?.includes('cancel')) {
-      responseMessage.text = 'Cancelled.';
-      return await this.handleResponseMessage(responseMessage, saveOptions, user);
-    }
-
-    // If error occurred during generation (likely token_balance)
-    if (this.result?.errorMessage?.length > 0) {
-      responseMessage.error = true;
-      responseMessage.text = this.result.output;
-      return await this.handleResponseMessage(responseMessage, saveOptions, user);
-    }
-
-    if (this.agentOptions.skipCompletion && this.result.output && this.functionsAgent) {
-      const partialText = opts.getPartialText();
-      const trimmedPartial = opts.getPartialText().replaceAll(':::plugin:::\n', '');
-      responseMessage.text =
-        trimmedPartial.length === 0 ? `${partialText}${this.result.output}` : partialText;
-      addImages(this.result.intermediateSteps, responseMessage);
-      await this.generateTextStream(this.result.output, opts.onProgress, { delay: 5 });
-      return await this.handleResponseMessage(responseMessage, saveOptions, user);
-    }
-
-    if (this.agentOptions.skipCompletion && this.result.output) {
-      responseMessage.text = this.result.output;
-      addImages(this.result.intermediateSteps, responseMessage);
-      await this.generateTextStream(this.result.output, opts.onProgress, { delay: 5 });
-      return await this.handleResponseMessage(responseMessage, saveOptions, user);
-    }
-
-    logger.debug('[PluginsClient] Completion phase: this.result', this.result);
-
-    const promptPrefix = buildPromptPrefix({
-      result: this.result,
-      message,
-      functionsAgent: this.functionsAgent,
-    });
-
-    logger.debug('[PluginsClient]', { promptPrefix });
-
-    payload = await this.buildCompletionPrompt({
-      messages: this.currentMessages,
-      promptPrefix,
-    });
-
-    logger.debug('[PluginsClient] buildCompletionPrompt Payload', payload);
-    responseMessage.text = await this.sendCompletion(payload, opts);
-    return await this.handleResponseMessage(responseMessage, saveOptions, user);
-  }
-
-  async buildCompletionPrompt({ messages, promptPrefix: _promptPrefix }) {
-    logger.debug('[PluginsClient] buildCompletionPrompt messages', messages);
-
-    const orderedMessages = messages;
-    let promptPrefix = _promptPrefix.trim();
-    // If the prompt prefix doesn't end with the end token, add it.
-    if (!promptPrefix.endsWith(`${this.endToken}`)) {
-      promptPrefix = `${promptPrefix.trim()}${this.endToken}\n\n`;
-    }
-    promptPrefix = `${this.startToken}Instructions:\n${promptPrefix}`;
-    const promptSuffix = `${this.startToken}${this.chatGptLabel ?? 'Assistant'}:\n`;
-
-    const instructionsPayload = {
-      role: 'system',
-      content: promptPrefix,
-    };
-
-    const messagePayload = {
-      role: 'system',
-      content: promptSuffix,
-    };
-
-    if (this.isGpt3) {
-      instructionsPayload.role = 'user';
-      messagePayload.role = 'user';
-      instructionsPayload.content += `\n${promptSuffix}`;
-    }
-
-    // testing if this works with browser endpoint
-    if (!this.isGpt3 && this.options.reverseProxyUrl) {
-      instructionsPayload.role = 'user';
-    }
-
-    let currentTokenCount =
-      this.getTokenCountForMessage(instructionsPayload) +
-      this.getTokenCountForMessage(messagePayload);
-
-    let promptBody = '';
-    const maxTokenCount = this.maxPromptTokens;
-    // Iterate backwards through the messages, adding them to the prompt until we reach the max token count.
-    // Do this within a recursive async function so that it doesn't block the event loop for too long.
-    const buildPromptBody = async () => {
-      if (currentTokenCount < maxTokenCount && orderedMessages.length > 0) {
-        const message = orderedMessages.pop();
-        const isCreatedByUser = message.isCreatedByUser || message.role?.toLowerCase() === 'user';
-        const roleLabel = isCreatedByUser ? this.userLabel : this.chatGptLabel;
-        let messageString = `${this.startToken}${roleLabel}:\n${
-          message.text ?? message.content ?? ''
-        }${this.endToken}\n`;
-        let newPromptBody = `${messageString}${promptBody}`;
-
-        const tokenCountForMessage = this.getTokenCount(messageString);
-        const newTokenCount = currentTokenCount + tokenCountForMessage;
-        if (newTokenCount > maxTokenCount) {
-          if (promptBody) {
-            // This message would put us over the token limit, so don't add it.
-            return false;
-          }
-          // This is the first message, so we can't add it. Just throw an error.
-          throw new Error(
-            `Prompt is too long. Max token count is ${maxTokenCount}, but prompt is ${newTokenCount} tokens long.`,
-          );
-        }
-        promptBody = newPromptBody;
-        currentTokenCount = newTokenCount;
-        // wait for next tick to avoid blocking the event loop
-        await new Promise((resolve) => setTimeout(resolve, 0));
-        return buildPromptBody();
-      }
-      return true;
-    };
-
-    await buildPromptBody();
-    const prompt = promptBody;
-    messagePayload.content = prompt;
-    // Add 2 tokens for metadata after all messages have been counted.
-    currentTokenCount += 2;
-
-    if (this.isGpt3 && messagePayload.content.length > 0) {
-      const context = 'Chat History:\n';
-      messagePayload.content = `${context}${prompt}`;
-      currentTokenCount += this.getTokenCount(context);
-    }
-
-    // Use up to `this.maxContextTokens` tokens (prompt + response), but try to leave `this.maxTokens` tokens for the response.
-    this.modelOptions.max_tokens = Math.min(
-      this.maxContextTokens - currentTokenCount,
-      this.maxResponseTokens,
-    );
-
-    if (this.isGpt3) {
-      messagePayload.content += promptSuffix;
-      return [instructionsPayload, messagePayload];
-    }
-
-    const result = [messagePayload, instructionsPayload];
-
-    if (this.functionsAgent && !this.isGpt3) {
-      result[1].content = `${result[1].content}\n${this.startToken}${this.chatGptLabel}:\nSure thing! Here is the output you requested:\n`;
-    }
-
-    return result.filter((message) => message.content.length > 0);
-  }
-}
-
-module.exports = PluginsClient;
--- a/api/app/clients/callbacks/createStartHandler.js
+++ b/api/app/clients/callbacks/createStartHandler.js
@@ -1,8 +1,8 @@
 const { promptTokensEstimate } = require('openai-chat-tokens');
 const { EModelEndpoint, supportsBalanceCheck } = require('librechat-data-provider');
 const { formatFromLangChain } = require('~/app/clients/prompts');
-const checkBalance = require('~/models/checkBalance');
-const { isEnabled } = require('~/server/utils');
+const { getBalanceConfig } = require('~/server/services/Config');
+const { checkBalance } = require('~/models/balanceMethods');
 const { logger } = require('~/config');

 const createStartHandler = ({
@@ -49,8 +49,8 @@ const createStartHandler = ({
    prelimPromptTokens += tokenBuffer;

    try {
-      // TODO: if plugins extends to non-OpenAI models, this will need to be updated
-      if (isEnabled(process.env.CHECK_BALANCE) && supportsBalanceCheck[EModelEndpoint.openAI]) {
+      const balance = await getBalanceConfig();
+      if (balance?.enabled && supportsBalanceCheck[EModelEndpoint.openAI]) {
        const generations =
          initialMessageCount && messages.length > initialMessageCount
            ? messages.slice(initialMessageCount)
--- a/api/app/clients/index.js
+++ b/api/app/clients/index.js
@@ -1,15 +1,11 @@
-const ChatGPTClient = require('./ChatGPTClient');
 const OpenAIClient = require('./OpenAIClient');
-const PluginsClient = require('./PluginsClient');
 const GoogleClient = require('./GoogleClient');
 const TextStream = require('./TextStream');
 const AnthropicClient = require('./AnthropicClient');
 const toolUtils = require('./tools/util');

 module.exports = {
-  ChatGPTClient,
  OpenAIClient,
-  PluginsClient,
  GoogleClient,
  TextStream,
  AnthropicClient,
--- a/api/app/clients/llm/createLLM.js
+++ b/api/app/clients/llm/createLLM.js
@@ -1,6 +1,5 @@
 const { ChatOpenAI } = require('@langchain/openai');
-const { sanitizeModelName, constructAzureURL } = require('~/utils');
-const { isEnabled } = require('~/server/utils');
+const { isEnabled, sanitizeModelName, constructAzureURL } = require('@librechat/api');

 /**
 * Creates a new instance of a language model (LLM) for chat interactions.
@@ -34,6 +33,7 @@ function createLLM({
  let credentials = { openAIApiKey };
  let configuration = {
    apiKey: openAIApiKey,
+    ...(configOptions.basePath && { baseURL: configOptions.basePath }),
  };

  /**  @type {AzureOptions} */
--- a/api/app/clients/prompts/addCacheControl.js
+++ b/api/app/clients/prompts/addCacheControl.js
@@ -1,7 +1,7 @@
 /**
 * Anthropic API: Adds cache control to the appropriate user messages in the payload.
- * @param {Array<AnthropicMessage>} messages - The array of message objects.
- * @returns {Array<AnthropicMessage>} - The updated array of message objects with cache control added.
+ * @param {Array<AnthropicMessage | BaseMessage>} messages - The array of message objects.
+ * @returns {Array<AnthropicMessage | BaseMessage>} - The updated array of message objects with cache control added.
 */
 function addCacheControl(messages) {
  if (!Array.isArray(messages) || messages.length < 2) {
@@ -13,7 +13,9 @@ function addCacheControl(messages) {

  for (let i = updatedMessages.length - 1; i >= 0 && userMessagesModified < 2; i--) {
    const message = updatedMessages[i];
-    if (message.role !== 'user') {
+    if (message.getType != null && message.getType() !== 'human') {
+      continue;
+    } else if (message.getType == null && message.role !== 'user') {
      continue;
    }

--- a/api/app/clients/prompts/createContextHandlers.js
+++ b/api/app/clients/prompts/createContextHandlers.js
@@ -1,6 +1,7 @@
 const axios = require('axios');
-const { isEnabled } = require('~/server/utils');
-const { logger } = require('~/config');
+const { isEnabled } = require('@librechat/api');
+const { logger } = require('@librechat/data-schemas');
+const { generateShortLivedToken } = require('~/server/services/AuthService');

 const footer = `Use the context as your learned knowledge to better answer the user.

@@ -18,7 +19,7 @@ function createContextHandlers(req, userMessageContent) {
  const queryPromises = [];
  const processedFiles = [];
  const processedIds = new Set();
-  const jwtToken = req.headers.authorization.split(' ')[1];
+  const jwtToken = generateShortLivedToken(req.user.id);
  const useFullContext = isEnabled(process.env.RAG_USE_FULL_CONTEXT);

  const query = async (file) => {
@@ -96,35 +97,35 @@ function createContextHandlers(req, userMessageContent) {
        resolvedQueries.length === 0
          ? '\n\tThe semantic search did not return any results.'
          : resolvedQueries
-            .map((queryResult, index) => {
-              const file = processedFiles[index];
-              let contextItems = queryResult.data;
+              .map((queryResult, index) => {
+                const file = processedFiles[index];
+                let contextItems = queryResult.data;

-              const generateContext = (currentContext) =>
-                `
+                const generateContext = (currentContext) =>
+                  `
          <file>
            <filename>${file.filename}</filename>
            <context>${currentContext}
            </context>
          </file>`;

-              if (useFullContext) {
-                return generateContext(`\n${contextItems}`);
-              }
+                if (useFullContext) {
+                  return generateContext(`\n${contextItems}`);
+                }

-              contextItems = queryResult.data
-                .map((item) => {
-                  const pageContent = item[0].page_content;
-                  return `
+                contextItems = queryResult.data
+                  .map((item) => {
+                    const pageContent = item[0].page_content;
+                    return `
            <contextItem>
              <![CDATA[${pageContent?.trim()}]]>
            </contextItem>`;
-                })
-                .join('');
+                  })
+                  .join('');

-              return generateContext(contextItems);
-            })
-            .join('');
+                return generateContext(contextItems);
+              })
+              .join('');

      if (useFullContext) {
        const prompt = `${header}
--- a/api/app/clients/prompts/formatAgentMessages.spec.js
+++ b/api/app/clients/prompts/formatAgentMessages.spec.js
@@ -325,4 +325,37 @@ describe('formatAgentMessages', () => {
    );
    expect(result[0].content).not.toContain('Analyzing the problem...');
  });
+
+  it('should exclude ERROR type content parts', () => {
+    const payload = [
+      {
+        role: 'assistant',
+        content: [
+          { type: ContentTypes.TEXT, [ContentTypes.TEXT]: 'Hello there' },
+          {
+            type: ContentTypes.ERROR,
+            [ContentTypes.ERROR]:
+              'An error occurred while processing the request: Something went wrong',
+          },
+          { type: ContentTypes.TEXT, [ContentTypes.TEXT]: 'Final answer' },
+        ],
+      },
+    ];
+
+    const result = formatAgentMessages(payload);
+
+    expect(result).toHaveLength(1);
+    expect(result[0]).toBeInstanceOf(AIMessage);
+    expect(result[0].content).toEqual([
+      { type: ContentTypes.TEXT, [ContentTypes.TEXT]: 'Hello there' },
+      { type: ContentTypes.TEXT, [ContentTypes.TEXT]: 'Final answer' },
+    ]);
+
+    // Make sure no error content exists in the result
+    const hasErrorContent = result[0].content.some(
+      (item) =>
+        item.type === ContentTypes.ERROR || JSON.stringify(item).includes('An error occurred'),
+    );
+    expect(hasErrorContent).toBe(false);
+  });
 });
--- a/api/app/clients/prompts/formatMessages.js
+++ b/api/app/clients/prompts/formatMessages.js
@@ -211,6 +211,8 @@ const formatAgentMessages = (payload) => {
      } else if (part.type === ContentTypes.THINK) {
        hasReasoning = true;
        continue;
+      } else if (part.type === ContentTypes.ERROR || part.type === ContentTypes.AGENT_UPDATE) {
+        continue;
      } else {
        currentContent.push(part);
      }
@@ -235,41 +237,9 @@ const formatAgentMessages = (payload) => {
  return messages;
 };

-/**
- * Formats an array of messages for LangChain, making sure all content fields are strings
- * @param {Array<(HumanMessage|AIMessage|SystemMessage|ToolMessage)>} payload - The array of messages to format.
- * @returns {Array<(HumanMessage|AIMessage|SystemMessage|ToolMessage)>} - The array of formatted LangChain messages, including ToolMessages for tool calls.
- */
-const formatContentStrings = (payload) => {
-  const messages = [];
-
-  for (const message of payload) {
-    if (typeof message.content === 'string') {
-      continue;
-    }
-
-    if (!Array.isArray(message.content)) {
-      continue;
-    }
-
-    // Reduce text types to a single string, ignore all other types
-    const content = message.content.reduce((acc, curr) => {
-      if (curr.type === ContentTypes.TEXT) {
-        return `${acc}${curr[ContentTypes.TEXT]}\n`;
-      }
-      return acc;
-    }, '');
-
-    message.content = content.trim();
-  }
-
-  return messages;
-};
-
 module.exports = {
  formatMessage,
  formatFromLangChain,
  formatAgentMessages,
-  formatContentStrings,
  formatLangChainMessages,
 };
--- a/api/app/clients/specs/AnthropicClient.test.js
+++ b/api/app/clients/specs/AnthropicClient.test.js
@@ -1,3 +1,4 @@
+const { SplitStreamHandler } = require('@librechat/agents');
 const { anthropicSettings } = require('librechat-data-provider');
 const AnthropicClient = require('~/app/clients/AnthropicClient');

@@ -14,7 +15,7 @@ describe('AnthropicClient', () => {
    {
      role: 'user',
      isCreatedByUser: true,
-      text: 'What\'s up',
+      text: "What's up",
      messageId: '3',
      parentMessageId: '2',
    },
@@ -169,7 +170,7 @@ describe('AnthropicClient', () => {
      client.options.modelLabel = 'Claude-2';
      const result = await client.buildMessages(messages, parentMessageId);
      const { prompt } = result;
-      expect(prompt).toContain('Human\'s name: John');
+      expect(prompt).toContain("Human's name: John");
      expect(prompt).toContain('You are Claude-2');
    });
  });
@@ -243,6 +244,64 @@ describe('AnthropicClient', () => {
      );
    });

+    describe('Claude 4 model headers', () => {
+      it('should add "prompt-caching" beta header for claude-sonnet-4 model', () => {
+        const client = new AnthropicClient('test-api-key');
+        const modelOptions = {
+          model: 'claude-sonnet-4-20250514',
+        };
+        client.setOptions({ modelOptions, promptCache: true });
+        const anthropicClient = client.getClient(modelOptions);
+        expect(anthropicClient._options.defaultHeaders).toBeDefined();
+        expect(anthropicClient._options.defaultHeaders).toHaveProperty('anthropic-beta');
+        expect(anthropicClient._options.defaultHeaders['anthropic-beta']).toBe(
+          'prompt-caching-2024-07-31',
+        );
+      });
+
+      it('should add "prompt-caching" beta header for claude-opus-4 model', () => {
+        const client = new AnthropicClient('test-api-key');
+        const modelOptions = {
+          model: 'claude-opus-4-20250514',
+        };
+        client.setOptions({ modelOptions, promptCache: true });
+        const anthropicClient = client.getClient(modelOptions);
+        expect(anthropicClient._options.defaultHeaders).toBeDefined();
+        expect(anthropicClient._options.defaultHeaders).toHaveProperty('anthropic-beta');
+        expect(anthropicClient._options.defaultHeaders['anthropic-beta']).toBe(
+          'prompt-caching-2024-07-31',
+        );
+      });
+
+      it('should add "prompt-caching" beta header for claude-4-sonnet model', () => {
+        const client = new AnthropicClient('test-api-key');
+        const modelOptions = {
+          model: 'claude-4-sonnet-20250514',
+        };
+        client.setOptions({ modelOptions, promptCache: true });
+        const anthropicClient = client.getClient(modelOptions);
+        expect(anthropicClient._options.defaultHeaders).toBeDefined();
+        expect(anthropicClient._options.defaultHeaders).toHaveProperty('anthropic-beta');
+        expect(anthropicClient._options.defaultHeaders['anthropic-beta']).toBe(
+          'prompt-caching-2024-07-31',
+        );
+      });
+
+      it('should add "prompt-caching" beta header for claude-4-opus model', () => {
+        const client = new AnthropicClient('test-api-key');
+        const modelOptions = {
+          model: 'claude-4-opus-20250514',
+        };
+        client.setOptions({ modelOptions, promptCache: true });
+        const anthropicClient = client.getClient(modelOptions);
+        expect(anthropicClient._options.defaultHeaders).toBeDefined();
+        expect(anthropicClient._options.defaultHeaders).toHaveProperty('anthropic-beta');
+        expect(anthropicClient._options.defaultHeaders['anthropic-beta']).toBe(
+          'prompt-caching-2024-07-31',
+        );
+      });
+    });
+
    it('should not add beta header for claude-3-5-sonnet-latest model', () => {
      const client = new AnthropicClient('test-api-key');
      const modelOptions = {
@@ -250,7 +309,7 @@ describe('AnthropicClient', () => {
      };
      client.setOptions({ modelOptions, promptCache: true });
      const anthropicClient = client.getClient(modelOptions);
-      expect(anthropicClient.defaultHeaders).not.toHaveProperty('anthropic-beta');
+      expect(anthropicClient._options.defaultHeaders).toBeUndefined();
    });

    it('should not add beta header for other models', () => {
@@ -261,7 +320,7 @@ describe('AnthropicClient', () => {
        },
      });
      const anthropicClient = client.getClient();
-      expect(anthropicClient.defaultHeaders).not.toHaveProperty('anthropic-beta');
+      expect(anthropicClient._options.defaultHeaders).toBeUndefined();
    });
  });

@@ -405,4 +464,574 @@ describe('AnthropicClient', () => {
      expect(Number.isNaN(result)).toBe(false);
    });
  });
+
+  describe('maxOutputTokens handling for different models', () => {
+    it('should not cap maxOutputTokens for Claude 3.5 Sonnet models', () => {
+      const client = new AnthropicClient('test-api-key');
+      const highTokenValue = anthropicSettings.legacy.maxOutputTokens.default * 10;
+
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3-5-sonnet',
+          maxOutputTokens: highTokenValue,
+        },
+      });
+
+      expect(client.modelOptions.maxOutputTokens).toBe(highTokenValue);
+
+      // Test with decimal notation
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3.5-sonnet',
+          maxOutputTokens: highTokenValue,
+        },
+      });
+
+      expect(client.modelOptions.maxOutputTokens).toBe(highTokenValue);
+    });
+
+    it('should not cap maxOutputTokens for Claude 3.7 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      const highTokenValue = anthropicSettings.legacy.maxOutputTokens.default * 2;
+
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3-7-sonnet',
+          maxOutputTokens: highTokenValue,
+        },
+      });
+
+      expect(client.modelOptions.maxOutputTokens).toBe(highTokenValue);
+
+      // Test with decimal notation
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3.7-sonnet',
+          maxOutputTokens: highTokenValue,
+        },
+      });
+
+      expect(client.modelOptions.maxOutputTokens).toBe(highTokenValue);
+    });
+
+    it('should not cap maxOutputTokens for Claude 4 Sonnet models', () => {
+      const client = new AnthropicClient('test-api-key');
+      const highTokenValue = anthropicSettings.legacy.maxOutputTokens.default * 10; // 40,960 tokens
+
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-sonnet-4-20250514',
+          maxOutputTokens: highTokenValue,
+        },
+      });
+
+      expect(client.modelOptions.maxOutputTokens).toBe(highTokenValue);
+    });
+
+    it('should not cap maxOutputTokens for Claude 4 Opus models', () => {
+      const client = new AnthropicClient('test-api-key');
+      const highTokenValue = anthropicSettings.legacy.maxOutputTokens.default * 6; // 24,576 tokens (under 32K limit)
+
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-opus-4-20250514',
+          maxOutputTokens: highTokenValue,
+        },
+      });
+
+      expect(client.modelOptions.maxOutputTokens).toBe(highTokenValue);
+    });
+
+    it('should cap maxOutputTokens for Claude 3.5 Haiku models', () => {
+      const client = new AnthropicClient('test-api-key');
+      const highTokenValue = anthropicSettings.legacy.maxOutputTokens.default * 2;
+
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3-5-haiku',
+          maxOutputTokens: highTokenValue,
+        },
+      });
+
+      expect(client.modelOptions.maxOutputTokens).toBe(
+        anthropicSettings.legacy.maxOutputTokens.default,
+      );
+
+      // Test with decimal notation
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3.5-haiku',
+          maxOutputTokens: highTokenValue,
+        },
+      });
+
+      expect(client.modelOptions.maxOutputTokens).toBe(
+        anthropicSettings.legacy.maxOutputTokens.default,
+      );
+    });
+
+    it('should cap maxOutputTokens for Claude 3 Haiku and Opus models', () => {
+      const client = new AnthropicClient('test-api-key');
+      const highTokenValue = anthropicSettings.legacy.maxOutputTokens.default * 2;
+
+      // Test haiku
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3-haiku',
+          maxOutputTokens: highTokenValue,
+        },
+      });
+
+      expect(client.modelOptions.maxOutputTokens).toBe(
+        anthropicSettings.legacy.maxOutputTokens.default,
+      );
+
+      // Test opus
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3-opus',
+          maxOutputTokens: highTokenValue,
+        },
+      });
+
+      expect(client.modelOptions.maxOutputTokens).toBe(
+        anthropicSettings.legacy.maxOutputTokens.default,
+      );
+    });
+  });
+
+  describe('topK/topP parameters for different models', () => {
+    beforeEach(() => {
+      // Mock the SplitStreamHandler
+      jest.spyOn(SplitStreamHandler.prototype, 'handle').mockImplementation(() => {});
+    });
+
+    afterEach(() => {
+      jest.restoreAllMocks();
+    });
+
+    it('should include top_k and top_p parameters for non-claude-3.7 models', async () => {
+      const client = new AnthropicClient('test-api-key');
+
+      // Create a mock async generator function
+      async function* mockAsyncGenerator() {
+        yield { type: 'message_start', message: { usage: {} } };
+        yield { delta: { text: 'Test response' } };
+        yield { type: 'message_delta', usage: {} };
+      }
+
+      // Mock createResponse to return the async generator
+      jest.spyOn(client, 'createResponse').mockImplementation(() => {
+        return mockAsyncGenerator();
+      });
+
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3-opus',
+          temperature: 0.7,
+          topK: 10,
+          topP: 0.9,
+        },
+      });
+
+      // Mock getClient to capture the request options
+      let capturedOptions = null;
+      jest.spyOn(client, 'getClient').mockImplementation((options) => {
+        capturedOptions = options;
+        return {};
+      });
+
+      const payload = [{ role: 'user', content: 'Test message' }];
+      await client.sendCompletion(payload, {});
+
+      // Check the options passed to getClient
+      expect(capturedOptions).toHaveProperty('top_k', 10);
+      expect(capturedOptions).toHaveProperty('top_p', 0.9);
+    });
+
+    it('should include top_k and top_p parameters for claude-3-5-sonnet models', async () => {
+      const client = new AnthropicClient('test-api-key');
+
+      // Create a mock async generator function
+      async function* mockAsyncGenerator() {
+        yield { type: 'message_start', message: { usage: {} } };
+        yield { delta: { text: 'Test response' } };
+        yield { type: 'message_delta', usage: {} };
+      }
+
+      // Mock createResponse to return the async generator
+      jest.spyOn(client, 'createResponse').mockImplementation(() => {
+        return mockAsyncGenerator();
+      });
+
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3-5-sonnet',
+          temperature: 0.7,
+          topK: 10,
+          topP: 0.9,
+        },
+      });
+
+      // Mock getClient to capture the request options
+      let capturedOptions = null;
+      jest.spyOn(client, 'getClient').mockImplementation((options) => {
+        capturedOptions = options;
+        return {};
+      });
+
+      const payload = [{ role: 'user', content: 'Test message' }];
+      await client.sendCompletion(payload, {});
+
+      // Check the options passed to getClient
+      expect(capturedOptions).toHaveProperty('top_k', 10);
+      expect(capturedOptions).toHaveProperty('top_p', 0.9);
+    });
+
+    it('should not include top_k and top_p parameters for claude-3-7-sonnet models', async () => {
+      const client = new AnthropicClient('test-api-key');
+
+      // Create a mock async generator function
+      async function* mockAsyncGenerator() {
+        yield { type: 'message_start', message: { usage: {} } };
+        yield { delta: { text: 'Test response' } };
+        yield { type: 'message_delta', usage: {} };
+      }
+
+      // Mock createResponse to return the async generator
+      jest.spyOn(client, 'createResponse').mockImplementation(() => {
+        return mockAsyncGenerator();
+      });
+
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3-7-sonnet',
+          temperature: 0.7,
+          topK: 10,
+          topP: 0.9,
+        },
+      });
+
+      // Mock getClient to capture the request options
+      let capturedOptions = null;
+      jest.spyOn(client, 'getClient').mockImplementation((options) => {
+        capturedOptions = options;
+        return {};
+      });
+
+      const payload = [{ role: 'user', content: 'Test message' }];
+      await client.sendCompletion(payload, {});
+
+      // Check the options passed to getClient
+      expect(capturedOptions).not.toHaveProperty('top_k');
+      expect(capturedOptions).not.toHaveProperty('top_p');
+    });
+
+    it('should not include top_k and top_p parameters for models with decimal notation (claude-3.7)', async () => {
+      const client = new AnthropicClient('test-api-key');
+
+      // Create a mock async generator function
+      async function* mockAsyncGenerator() {
+        yield { type: 'message_start', message: { usage: {} } };
+        yield { delta: { text: 'Test response' } };
+        yield { type: 'message_delta', usage: {} };
+      }
+
+      // Mock createResponse to return the async generator
+      jest.spyOn(client, 'createResponse').mockImplementation(() => {
+        return mockAsyncGenerator();
+      });
+
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3.7-sonnet',
+          temperature: 0.7,
+          topK: 10,
+          topP: 0.9,
+        },
+      });
+
+      // Mock getClient to capture the request options
+      let capturedOptions = null;
+      jest.spyOn(client, 'getClient').mockImplementation((options) => {
+        capturedOptions = options;
+        return {};
+      });
+
+      const payload = [{ role: 'user', content: 'Test message' }];
+      await client.sendCompletion(payload, {});
+
+      // Check the options passed to getClient
+      expect(capturedOptions).not.toHaveProperty('top_k');
+      expect(capturedOptions).not.toHaveProperty('top_p');
+    });
+  });
+
+  it('should include top_k and top_p parameters for Claude-3.7 models when thinking is explicitly disabled', async () => {
+    const client = new AnthropicClient('test-api-key', {
+      modelOptions: {
+        model: 'claude-3-7-sonnet',
+        temperature: 0.7,
+        topK: 10,
+        topP: 0.9,
+      },
+      thinking: false,
+    });
+
+    async function* mockAsyncGenerator() {
+      yield { type: 'message_start', message: { usage: {} } };
+      yield { delta: { text: 'Test response' } };
+      yield { type: 'message_delta', usage: {} };
+    }
+
+    jest.spyOn(client, 'createResponse').mockImplementation(() => {
+      return mockAsyncGenerator();
+    });
+
+    let capturedOptions = null;
+    jest.spyOn(client, 'getClient').mockImplementation((options) => {
+      capturedOptions = options;
+      return {};
+    });
+
+    const payload = [{ role: 'user', content: 'Test message' }];
+    await client.sendCompletion(payload, {});
+
+    expect(capturedOptions).toHaveProperty('topK', 10);
+    expect(capturedOptions).toHaveProperty('topP', 0.9);
+
+    client.setOptions({
+      modelOptions: {
+        model: 'claude-3.7-sonnet',
+        temperature: 0.7,
+        topK: 10,
+        topP: 0.9,
+      },
+      thinking: false,
+    });
+
+    await client.sendCompletion(payload, {});
+
+    expect(capturedOptions).toHaveProperty('topK', 10);
+    expect(capturedOptions).toHaveProperty('topP', 0.9);
+  });
+
+  describe('isClaudeLatest', () => {
+    it('should set isClaudeLatest to true for claude-3 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3-sonnet-20240229',
+        },
+      });
+      expect(client.isClaudeLatest).toBe(true);
+    });
+
+    it('should set isClaudeLatest to true for claude-3.5 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3.5-sonnet-20240229',
+        },
+      });
+      expect(client.isClaudeLatest).toBe(true);
+    });
+
+    it('should set isClaudeLatest to true for claude-sonnet-4 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-sonnet-4-20240229',
+        },
+      });
+      expect(client.isClaudeLatest).toBe(true);
+    });
+
+    it('should set isClaudeLatest to true for claude-opus-4 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-opus-4-20240229',
+        },
+      });
+      expect(client.isClaudeLatest).toBe(true);
+    });
+
+    it('should set isClaudeLatest to true for claude-3.5-haiku models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3.5-haiku-20240229',
+        },
+      });
+      expect(client.isClaudeLatest).toBe(true);
+    });
+
+    it('should set isClaudeLatest to false for claude-2 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-2',
+        },
+      });
+      expect(client.isClaudeLatest).toBe(false);
+    });
+
+    it('should set isClaudeLatest to false for claude-instant models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-instant',
+        },
+      });
+      expect(client.isClaudeLatest).toBe(false);
+    });
+
+    it('should set isClaudeLatest to false for claude-sonnet-3 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-sonnet-3-20240229',
+        },
+      });
+      expect(client.isClaudeLatest).toBe(false);
+    });
+
+    it('should set isClaudeLatest to false for claude-opus-3 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-opus-3-20240229',
+        },
+      });
+      expect(client.isClaudeLatest).toBe(false);
+    });
+
+    it('should set isClaudeLatest to false for claude-haiku-3 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-haiku-3-20240229',
+        },
+      });
+      expect(client.isClaudeLatest).toBe(false);
+    });
+  });
+
+  describe('configureReasoning', () => {
+    it('should enable thinking for claude-opus-4 and claude-sonnet-4 models', async () => {
+      const client = new AnthropicClient('test-api-key');
+      // Create a mock async generator function
+      async function* mockAsyncGenerator() {
+        yield { type: 'message_start', message: { usage: {} } };
+        yield { delta: { text: 'Test response' } };
+        yield { type: 'message_delta', usage: {} };
+      }
+
+      // Mock createResponse to return the async generator
+      jest.spyOn(client, 'createResponse').mockImplementation(() => {
+        return mockAsyncGenerator();
+      });
+
+      // Test claude-opus-4
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-opus-4-20250514',
+        },
+        thinking: true,
+        thinkingBudget: 2000,
+      });
+
+      let capturedOptions = null;
+      jest.spyOn(client, 'getClient').mockImplementation((options) => {
+        capturedOptions = options;
+        return {};
+      });
+
+      const payload = [{ role: 'user', content: 'Test message' }];
+      await client.sendCompletion(payload, {});
+
+      expect(capturedOptions).toHaveProperty('thinking');
+      expect(capturedOptions.thinking).toEqual({
+        type: 'enabled',
+        budget_tokens: 2000,
+      });
+
+      // Test claude-sonnet-4
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-sonnet-4-20250514',
+        },
+        thinking: true,
+        thinkingBudget: 2000,
+      });
+
+      await client.sendCompletion(payload, {});
+
+      expect(capturedOptions).toHaveProperty('thinking');
+      expect(capturedOptions.thinking).toEqual({
+        type: 'enabled',
+        budget_tokens: 2000,
+      });
+    });
+  });
+});
+
+describe('Claude Model Tests', () => {
+  it('should handle Claude 3 and 4 series models correctly', () => {
+    const client = new AnthropicClient('test-key');
+    // Claude 3 series models
+    const claude3Models = [
+      'claude-3-opus-20240229',
+      'claude-3-sonnet-20240229',
+      'claude-3-haiku-20240307',
+      'claude-3-5-sonnet-20240620',
+      'claude-3-5-haiku-20240620',
+      'claude-3.5-sonnet-20240620',
+      'claude-3.5-haiku-20240620',
+      'claude-3.7-sonnet-20240620',
+      'claude-3.7-haiku-20240620',
+      'anthropic/claude-3-opus-20240229',
+      'claude-3-opus-20240229/anthropic',
+    ];
+
+    // Claude 4 series models
+    const claude4Models = [
+      'claude-sonnet-4-20250514',
+      'claude-opus-4-20250514',
+      'claude-4-sonnet-20250514',
+      'claude-4-opus-20250514',
+      'anthropic/claude-sonnet-4-20250514',
+      'claude-sonnet-4-20250514/anthropic',
+    ];
+
+    // Test Claude 3 series
+    claude3Models.forEach((model) => {
+      client.setOptions({ modelOptions: { model } });
+      expect(
+        /claude-[3-9]/.test(client.modelOptions.model) ||
+          /claude-(?:sonnet|opus|haiku)-[4-9]/.test(client.modelOptions.model),
+      ).toBe(true);
+    });
+
+    // Test Claude 4 series
+    claude4Models.forEach((model) => {
+      client.setOptions({ modelOptions: { model } });
+      expect(
+        /claude-[3-9]/.test(client.modelOptions.model) ||
+          /claude-(?:sonnet|opus|haiku)-[4-9]/.test(client.modelOptions.model),
+      ).toBe(true);
+    });
+
+    // Test non-Claude 3/4 models
+    const nonClaudeModels = ['claude-2', 'claude-instant', 'gpt-4', 'gpt-3.5-turbo'];
+
+    nonClaudeModels.forEach((model) => {
+      client.setOptions({ modelOptions: { model } });
+      expect(
+        /claude-[3-9]/.test(client.modelOptions.model) ||
+          /claude-(?:sonnet|opus|haiku)-[4-9]/.test(client.modelOptions.model),
+      ).toBe(false);
+    });
+  });
 });
--- a/api/app/clients/specs/BaseClient.test.js
+++ b/api/app/clients/specs/BaseClient.test.js
@@ -1,7 +1,7 @@
 const { Constants } = require('librechat-data-provider');
 const { initializeFakeClient } = require('./FakeClient');

-jest.mock('~/lib/db/connectDb');
+jest.mock('~/db/connect');
 jest.mock('~/models', () => ({
  User: jest.fn(),
  Key: jest.fn(),
@@ -30,8 +30,12 @@ jest.mock('~/models', () => ({
  updateFileUsage: jest.fn(),
 }));

-jest.mock('@langchain/openai', () => {
+const { getConvo, saveConvo } = require('~/models');
+
+jest.mock('@librechat/agents', () => {
+  const { Providers } = jest.requireActual('@librechat/agents');
  return {
+    Providers,
    ChatOpenAI: jest.fn().mockImplementation(() => {
      return {};
    }),
@@ -50,7 +54,7 @@ const messageHistory = [
  {
    role: 'user',
    isCreatedByUser: true,
-    text: 'What\'s up',
+    text: "What's up",
    messageId: '3',
    parentMessageId: '2',
  },
@@ -162,7 +166,7 @@ describe('BaseClient', () => {
    const result = await TestClient.getMessagesWithinTokenLimit({ messages });

    expect(result.context).toEqual(expectedContext);
-    expect(result.summaryIndex).toEqual(expectedIndex);
+    expect(result.messagesToRefine.length - 1).toEqual(expectedIndex);
    expect(result.remainingContextTokens).toBe(expectedRemainingContextTokens);
    expect(result.messagesToRefine).toEqual(expectedMessagesToRefine);
  });
@@ -198,7 +202,7 @@ describe('BaseClient', () => {
    const result = await TestClient.getMessagesWithinTokenLimit({ messages });

    expect(result.context).toEqual(expectedContext);
-    expect(result.summaryIndex).toEqual(expectedIndex);
+    expect(result.messagesToRefine.length - 1).toEqual(expectedIndex);
    expect(result.remainingContextTokens).toBe(expectedRemainingContextTokens);
    expect(result.messagesToRefine).toEqual(expectedMessagesToRefine);
  });
@@ -418,6 +422,46 @@ describe('BaseClient', () => {
      expect(response).toEqual(expectedResult);
    });

+    test('should replace responseMessageId with new UUID when isRegenerate is true and messageId ends with underscore', async () => {
+      const mockCrypto = require('crypto');
+      const newUUID = 'new-uuid-1234';
+      jest.spyOn(mockCrypto, 'randomUUID').mockReturnValue(newUUID);
+
+      const opts = {
+        isRegenerate: true,
+        responseMessageId: 'existing-message-id_',
+      };
+
+      await TestClient.setMessageOptions(opts);
+
+      expect(TestClient.responseMessageId).toBe(newUUID);
+      expect(TestClient.responseMessageId).not.toBe('existing-message-id_');
+
+      mockCrypto.randomUUID.mockRestore();
+    });
+
+    test('should not replace responseMessageId when isRegenerate is false', async () => {
+      const opts = {
+        isRegenerate: false,
+        responseMessageId: 'existing-message-id_',
+      };
+
+      await TestClient.setMessageOptions(opts);
+
+      expect(TestClient.responseMessageId).toBe('existing-message-id_');
+    });
+
+    test('should not replace responseMessageId when it does not end with underscore', async () => {
+      const opts = {
+        isRegenerate: true,
+        responseMessageId: 'existing-message-id',
+      };
+
+      await TestClient.setMessageOptions(opts);
+
+      expect(TestClient.responseMessageId).toBe('existing-message-id');
+    });
+
    test('sendMessage should work with provided conversationId and parentMessageId', async () => {
      const userMessage = 'Second message in the conversation';
      const opts = {
@@ -454,7 +498,7 @@ describe('BaseClient', () => {

      const chatMessages2 = await TestClient.loadHistory(conversationId, '3');
      expect(TestClient.currentMessages).toHaveLength(3);
-      expect(chatMessages2[chatMessages2.length - 1].text).toEqual('What\'s up');
+      expect(chatMessages2[chatMessages2.length - 1].text).toEqual("What's up");
    });

    /* Most of the new sendMessage logic revolving around edited/continued AI messages
@@ -540,10 +584,11 @@ describe('BaseClient', () => {

    test('saveMessageToDatabase is called with the correct arguments', async () => {
      const saveOptions = TestClient.getSaveOptions();
-      const user = {}; // Mock user
+      const user = {};
      const opts = { user };
+      const saveSpy = jest.spyOn(TestClient, 'saveMessageToDatabase');
      await TestClient.sendMessage('Hello, world!', opts);
-      expect(TestClient.saveMessageToDatabase).toHaveBeenCalledWith(
+      expect(saveSpy).toHaveBeenCalledWith(
        expect.objectContaining({
          sender: expect.any(String),
          text: expect.any(String),
@@ -557,6 +602,157 @@ describe('BaseClient', () => {
      );
    });

+    test('should handle existing conversation when getConvo retrieves one', async () => {
+      const existingConvo = {
+        conversationId: 'existing-convo-id',
+        endpoint: 'openai',
+        endpointType: 'openai',
+        model: 'gpt-3.5-turbo',
+        messages: [
+          { role: 'user', content: 'Existing message 1' },
+          { role: 'assistant', content: 'Existing response 1' },
+        ],
+        temperature: 1,
+      };
+
+      const { temperature: _temp, ...newConvo } = existingConvo;
+
+      const user = {
+        id: 'user-id',
+      };
+
+      getConvo.mockResolvedValue(existingConvo);
+      saveConvo.mockResolvedValue(newConvo);
+
+      TestClient = initializeFakeClient(
+        apiKey,
+        {
+          ...options,
+          req: {
+            user,
+          },
+        },
+        [],
+      );
+
+      const saveSpy = jest.spyOn(TestClient, 'saveMessageToDatabase');
+
+      const newMessage = 'New message in existing conversation';
+      const response = await TestClient.sendMessage(newMessage, {
+        user,
+        conversationId: existingConvo.conversationId,
+      });
+
+      expect(getConvo).toHaveBeenCalledWith(user.id, existingConvo.conversationId);
+      expect(TestClient.conversationId).toBe(existingConvo.conversationId);
+      expect(response.conversationId).toBe(existingConvo.conversationId);
+      expect(TestClient.fetchedConvo).toBe(true);
+
+      expect(saveSpy).toHaveBeenCalledWith(
+        expect.objectContaining({
+          conversationId: existingConvo.conversationId,
+          text: newMessage,
+        }),
+        expect.any(Object),
+        expect.any(Object),
+      );
+
+      expect(saveConvo).toHaveBeenCalledTimes(2);
+      expect(saveConvo).toHaveBeenCalledWith(
+        expect.any(Object),
+        expect.objectContaining({
+          conversationId: existingConvo.conversationId,
+        }),
+        expect.objectContaining({
+          context: 'api/app/clients/BaseClient.js - saveMessageToDatabase #saveConvo',
+          unsetFields: {
+            temperature: 1,
+          },
+        }),
+      );
+
+      await TestClient.sendMessage('Another message', {
+        conversationId: existingConvo.conversationId,
+      });
+      expect(getConvo).toHaveBeenCalledTimes(1);
+    });
+
+    test('should correctly handle existing conversation and unset fields appropriately', async () => {
+      const existingConvo = {
+        conversationId: 'existing-convo-id',
+        endpoint: 'openai',
+        endpointType: 'openai',
+        model: 'gpt-3.5-turbo',
+        messages: [
+          { role: 'user', content: 'Existing message 1' },
+          { role: 'assistant', content: 'Existing response 1' },
+        ],
+        title: 'Existing Conversation',
+        someExistingField: 'existingValue',
+        anotherExistingField: 'anotherValue',
+        temperature: 0.7,
+        modelLabel: 'GPT-3.5',
+      };
+
+      getConvo.mockResolvedValue(existingConvo);
+      saveConvo.mockResolvedValue(existingConvo);
+
+      TestClient = initializeFakeClient(
+        apiKey,
+        {
+          ...options,
+          modelOptions: {
+            model: 'gpt-4',
+            temperature: 0.5,
+          },
+        },
+        [],
+      );
+
+      const newMessage = 'New message in existing conversation';
+      await TestClient.sendMessage(newMessage, {
+        conversationId: existingConvo.conversationId,
+      });
+
+      expect(saveConvo).toHaveBeenCalledTimes(2);
+
+      const saveConvoCall = saveConvo.mock.calls[0];
+      const [, savedFields, saveOptions] = saveConvoCall;
+
+      // Instead of checking all excludedKeys, we'll just check specific fields
+      // that we know should be excluded
+      expect(savedFields).not.toHaveProperty('messages');
+      expect(savedFields).not.toHaveProperty('title');
+
+      // Only check that someExistingField is in unsetFields
+      expect(saveOptions.unsetFields).toHaveProperty('someExistingField', 1);
+
+      // Mock saveConvo to return the expected fields
+      saveConvo.mockImplementation((req, fields) => {
+        return Promise.resolve({
+          ...fields,
+          endpoint: 'openai',
+          endpointType: 'openai',
+          model: 'gpt-4',
+          temperature: 0.5,
+        });
+      });
+
+      // Only check the conversationId since that's the only field we can be sure about
+      expect(savedFields).toHaveProperty('conversationId', 'existing-convo-id');
+
+      expect(TestClient.fetchedConvo).toBe(true);
+
+      await TestClient.sendMessage('Another message', {
+        conversationId: existingConvo.conversationId,
+      });
+
+      expect(getConvo).toHaveBeenCalledTimes(1);
+
+      const secondSaveConvoCall = saveConvo.mock.calls[1];
+      expect(secondSaveConvoCall[2]).toHaveProperty('unsetFields', {});
+    });
+
    test('sendCompletion is called with the correct arguments', async () => {
      const payload = {}; // Mock payload
      TestClient.buildMessages.mockReturnValue({ prompt: payload, tokenCountMap: null });
--- a/api/app/clients/specs/FakeClient.js
+++ b/api/app/clients/specs/FakeClient.js
@@ -56,7 +56,6 @@ const initializeFakeClient = (apiKey, options, fakeMessages) => {
  let TestClient = new FakeClient(apiKey);
  TestClient.options = options;
  TestClient.abortController = { abort: jest.fn() };
-  TestClient.saveMessageToDatabase = jest.fn();
  TestClient.loadHistory = jest
    .fn()
    .mockImplementation((conversationId, parentMessageId = null) => {
@@ -86,7 +85,6 @@ const initializeFakeClient = (apiKey, options, fakeMessages) => {
    return 'Mock response text';
  });

-  // eslint-disable-next-line no-unused-vars
  TestClient.getCompletion = jest.fn().mockImplementation(async (..._args) => {
    return {
      choices: [
--- a/api/app/clients/specs/OpenAIClient.test.js
+++ b/api/app/clients/specs/OpenAIClient.test.js
@@ -1,13 +1,11 @@
 jest.mock('~/cache/getLogStores');
 require('dotenv').config();
-const OpenAI = require('openai');
-const getLogStores = require('~/cache/getLogStores');
 const { fetchEventSource } = require('@waylaidwanderer/fetch-event-source');
-const { genAzureChatCompletion } = require('~/utils/azureUtils');
+const getLogStores = require('~/cache/getLogStores');
 const OpenAIClient = require('../OpenAIClient');
 jest.mock('meilisearch');

-jest.mock('~/lib/db/connectDb');
+jest.mock('~/db/connect');
 jest.mock('~/models', () => ({
  User: jest.fn(),
  Key: jest.fn(),
@@ -36,19 +34,21 @@ jest.mock('~/models', () => ({
  updateFileUsage: jest.fn(),
 }));

-jest.mock('@langchain/openai', () => {
-  return {
-    ChatOpenAI: jest.fn().mockImplementation(() => {
-      return {};
-    }),
-  };
+// Import the actual module but mock specific parts
+const agents = jest.requireActual('@librechat/agents');
+const { CustomOpenAIClient } = agents;
+
+// Also mock ChatOpenAI to prevent real API calls
+agents.ChatOpenAI = jest.fn().mockImplementation(() => {
+  return {};
+});
+agents.AzureChatOpenAI = jest.fn().mockImplementation(() => {
+  return {};
 });

-jest.mock('openai');
-
-jest.spyOn(OpenAI, 'constructor').mockImplementation(function (...options) {
-  // We can add additional logic here if needed
-  return new OpenAI(...options);
+// Mock only the CustomOpenAIClient constructor
+jest.spyOn(CustomOpenAIClient, 'constructor').mockImplementation(function (...options) {
+  return new CustomOpenAIClient(...options);
 });

 const finalChatCompletion = jest.fn().mockResolvedValue({
@@ -120,7 +120,13 @@ const create = jest.fn().mockResolvedValue({
  ],
 });

-OpenAI.mockImplementation(() => ({
+// Mock the implementation of CustomOpenAIClient instances
+jest.spyOn(CustomOpenAIClient.prototype, 'constructor').mockImplementation(function () {
+  return this;
+});
+
+// Create a mock for the CustomOpenAIClient class
+const mockCustomOpenAIClient = jest.fn().mockImplementation(() => ({
  beta: {
    chat: {
      completions: {
@@ -135,11 +141,14 @@ OpenAI.mockImplementation(() => ({
  },
 }));

-describe('OpenAIClient', () => {
-  const mockSet = jest.fn();
-  const mockCache = { set: mockSet };
+CustomOpenAIClient.mockImplementation = mockCustomOpenAIClient;

+describe('OpenAIClient', () => {
  beforeEach(() => {
+    const mockCache = {
+      get: jest.fn().mockResolvedValue({}),
+      set: jest.fn(),
+    };
    getLogStores.mockReturnValue(mockCache);
  });
  let client;
@@ -202,14 +211,6 @@ describe('OpenAIClient', () => {
      expect(client.modelOptions.temperature).toBe(0.7);
    });

-    it('should set apiKey and useOpenRouter if OPENROUTER_API_KEY is present', () => {
-      process.env.OPENROUTER_API_KEY = 'openrouter-key';
-      client.setOptions({});
-      expect(client.apiKey).toBe('openrouter-key');
-      expect(client.useOpenRouter).toBe(true);
-      delete process.env.OPENROUTER_API_KEY; // Cleanup
-    });
-
    it('should set FORCE_PROMPT based on OPENAI_FORCE_PROMPT or reverseProxyUrl', () => {
      process.env.OPENAI_FORCE_PROMPT = 'true';
      client.setOptions({});
@@ -461,17 +462,17 @@ describe('OpenAIClient', () => {
        role: 'system',
        name: 'example_user',
        content:
-          'Let\'s circle back when we have more bandwidth to touch base on opportunities for increased leverage.',
+          "Let's circle back when we have more bandwidth to touch base on opportunities for increased leverage.",
      },
      {
        role: 'system',
        name: 'example_assistant',
-        content: 'Let\'s talk later when we\'re less busy about how to do better.',
+        content: "Let's talk later when we're less busy about how to do better.",
      },
      {
        role: 'user',
        content:
-          'This late pivot means we don\'t have time to boil the ocean for the client deliverable.',
+          "This late pivot means we don't have time to boil the ocean for the client deliverable.",
      },
    ];

@@ -530,80 +531,6 @@ describe('OpenAIClient', () => {
    });
  });

-  describe('sendMessage/getCompletion/chatCompletion', () => {
-    afterEach(() => {
-      delete process.env.AZURE_OPENAI_DEFAULT_MODEL;
-      delete process.env.AZURE_USE_MODEL_AS_DEPLOYMENT_NAME;
-      delete process.env.OPENROUTER_API_KEY;
-    });
-
-    it('should call getCompletion and fetchEventSource when using a text/instruct model', async () => {
-      const model = 'text-davinci-003';
-      const onProgress = jest.fn().mockImplementation(() => ({}));
-
-      const testClient = new OpenAIClient('test-api-key', {
-        ...defaultOptions,
-        modelOptions: { model },
-      });
-
-      const getCompletion = jest.spyOn(testClient, 'getCompletion');
-      await testClient.sendMessage('Hi mom!', { onProgress });
-
-      expect(getCompletion).toHaveBeenCalled();
-      expect(getCompletion.mock.calls.length).toBe(1);
-
-      expect(getCompletion.mock.calls[0][0]).toBe('||>User:\nHi mom!\n||>Assistant:\n');
-
-      expect(fetchEventSource).toHaveBeenCalled();
-      expect(fetchEventSource.mock.calls.length).toBe(1);
-
-      // Check if the first argument (url) is correct
-      const firstCallArgs = fetchEventSource.mock.calls[0];
-
-      const expectedURL = 'https://api.openai.com/v1/completions';
-      expect(firstCallArgs[0]).toBe(expectedURL);
-
-      const requestBody = JSON.parse(firstCallArgs[1].body);
-      expect(requestBody).toHaveProperty('model');
-      expect(requestBody.model).toBe(model);
-    });
-
-    it('[Azure OpenAI] should call chatCompletion and OpenAI.stream with correct args', async () => {
-      // Set a default model
-      process.env.AZURE_OPENAI_DEFAULT_MODEL = 'gpt4-turbo';
-
-      const onProgress = jest.fn().mockImplementation(() => ({}));
-      client.azure = defaultAzureOptions;
-      const chatCompletion = jest.spyOn(client, 'chatCompletion');
-      await client.sendMessage('Hi mom!', {
-        replaceOptions: true,
-        ...defaultOptions,
-        modelOptions: { model: 'gpt4-turbo', stream: true },
-        onProgress,
-        azure: defaultAzureOptions,
-      });
-
-      expect(chatCompletion).toHaveBeenCalled();
-      expect(chatCompletion.mock.calls.length).toBe(1);
-
-      const chatCompletionArgs = chatCompletion.mock.calls[0][0];
-      const { payload } = chatCompletionArgs;
-
-      expect(payload[0].role).toBe('user');
-      expect(payload[0].content).toBe('Hi mom!');
-
-      // Azure OpenAI does not use the model property, and will error if it's passed
-      // This check ensures the model property is not present
-      const streamArgs = stream.mock.calls[0][0];
-      expect(streamArgs).not.toHaveProperty('model');
-
-      // Check if the baseURL is correct
-      const constructorArgs = OpenAI.mock.calls[0][0];
-      const expectedURL = genAzureChatCompletion(defaultAzureOptions).split('/chat')[0];
-      expect(constructorArgs.baseURL).toBe(expectedURL);
-    });
-  });
-
  describe('checkVisionRequest functionality', () => {
    let client;
    const attachments = [{ type: 'image/png' }];
--- a/api/app/clients/specs/PluginsClient.test.js
+++ b/api/app/clients/specs/PluginsClient.test.js
@@ -1,314 +0,0 @@
-const crypto = require('crypto');
-const { Constants } = require('librechat-data-provider');
-const { HumanMessage, AIMessage } = require('@langchain/core/messages');
-const PluginsClient = require('../PluginsClient');
-
-jest.mock('~/lib/db/connectDb');
-jest.mock('~/models/Conversation', () => {
-  return function () {
-    return {
-      save: jest.fn(),
-      deleteConvos: jest.fn(),
-    };
-  };
-});
-
-const defaultAzureOptions = {
-  azureOpenAIApiInstanceName: 'your-instance-name',
-  azureOpenAIApiDeploymentName: 'your-deployment-name',
-  azureOpenAIApiVersion: '2020-07-01-preview',
-};
-
-describe('PluginsClient', () => {
-  let TestAgent;
-  let options = {
-    tools: [],
-    modelOptions: {
-      model: 'gpt-3.5-turbo',
-      temperature: 0,
-      max_tokens: 2,
-    },
-    agentOptions: {
-      model: 'gpt-3.5-turbo',
-    },
-  };
-  let parentMessageId;
-  let conversationId;
-  const fakeMessages = [];
-  const userMessage = 'Hello, ChatGPT!';
-  const apiKey = 'fake-api-key';
-
-  beforeEach(() => {
-    TestAgent = new PluginsClient(apiKey, options);
-    TestAgent.loadHistory = jest
-      .fn()
-      .mockImplementation((conversationId, parentMessageId = null) => {
-        if (!conversationId) {
-          TestAgent.currentMessages = [];
-          return Promise.resolve([]);
-        }
-
-        const orderedMessages = TestAgent.constructor.getMessagesForConversation({
-          messages: fakeMessages,
-          parentMessageId,
-        });
-
-        const chatMessages = orderedMessages.map((msg) =>
-          msg?.isCreatedByUser || msg?.role?.toLowerCase() === 'user'
-            ? new HumanMessage(msg.text)
-            : new AIMessage(msg.text),
-        );
-
-        TestAgent.currentMessages = orderedMessages;
-        return Promise.resolve(chatMessages);
-      });
-    TestAgent.sendMessage = jest.fn().mockImplementation(async (message, opts = {}) => {
-      if (opts && typeof opts === 'object') {
-        TestAgent.setOptions(opts);
-      }
-      const conversationId = opts.conversationId || crypto.randomUUID();
-      const parentMessageId = opts.parentMessageId || Constants.NO_PARENT;
-      const userMessageId = opts.overrideParentMessageId || crypto.randomUUID();
-      this.pastMessages = await TestAgent.loadHistory(
-        conversationId,
-        TestAgent.options?.parentMessageId,
-      );
-
-      const userMessage = {
-        text: message,
-        sender: 'ChatGPT',
-        isCreatedByUser: true,
-        messageId: userMessageId,
-        parentMessageId,
-        conversationId,
-      };
-
-      const response = {
-        sender: 'ChatGPT',
-        text: 'Hello, User!',
-        isCreatedByUser: false,
-        messageId: crypto.randomUUID(),
-        parentMessageId: userMessage.messageId,
-        conversationId,
-      };
-
-      fakeMessages.push(userMessage);
-      fakeMessages.push(response);
-      return response;
-    });
-  });
-
-  test('initializes PluginsClient without crashing', () => {
-    expect(TestAgent).toBeInstanceOf(PluginsClient);
-  });
-
-  test('check setOptions function', () => {
-    expect(TestAgent.agentIsGpt3).toBe(true);
-  });
-
-  describe('sendMessage', () => {
-    test('sendMessage should return a response message', async () => {
-      const expectedResult = expect.objectContaining({
-        sender: 'ChatGPT',
-        text: expect.any(String),
-        isCreatedByUser: false,
-        messageId: expect.any(String),
-        parentMessageId: expect.any(String),
-        conversationId: expect.any(String),
-      });
-
-      const response = await TestAgent.sendMessage(userMessage);
-      parentMessageId = response.messageId;
-      conversationId = response.conversationId;
-      expect(response).toEqual(expectedResult);
-    });
-
-    test('sendMessage should work with provided conversationId and parentMessageId', async () => {
-      const userMessage = 'Second message in the conversation';
-      const opts = {
-        conversationId,
-        parentMessageId,
-      };
-
-      const expectedResult = expect.objectContaining({
-        sender: 'ChatGPT',
-        text: expect.any(String),
-        isCreatedByUser: false,
-        messageId: expect.any(String),
-        parentMessageId: expect.any(String),
-        conversationId: opts.conversationId,
-      });
-
-      const response = await TestAgent.sendMessage(userMessage, opts);
-      parentMessageId = response.messageId;
-      expect(response.conversationId).toEqual(conversationId);
-      expect(response).toEqual(expectedResult);
-    });
-
-    test('should return chat history', async () => {
-      const chatMessages = await TestAgent.loadHistory(conversationId, parentMessageId);
-      expect(TestAgent.currentMessages).toHaveLength(4);
-      expect(chatMessages[0].text).toEqual(userMessage);
-    });
-  });
-
-  describe('getFunctionModelName', () => {
-    let client;
-
-    beforeEach(() => {
-      client = new PluginsClient('dummy_api_key');
-    });
-
-    test('should return the input when it includes a dash followed by four digits', () => {
-      expect(client.getFunctionModelName('-1234')).toBe('-1234');
-      expect(client.getFunctionModelName('gpt-4-5678-preview')).toBe('gpt-4-5678-preview');
-    });
-
-    test('should return the input for all function-capable models (`0613` models and above)', () => {
-      expect(client.getFunctionModelName('gpt-4-0613')).toBe('gpt-4-0613');
-      expect(client.getFunctionModelName('gpt-4-32k-0613')).toBe('gpt-4-32k-0613');
-      expect(client.getFunctionModelName('gpt-3.5-turbo-0613')).toBe('gpt-3.5-turbo-0613');
-      expect(client.getFunctionModelName('gpt-3.5-turbo-16k-0613')).toBe('gpt-3.5-turbo-16k-0613');
-      expect(client.getFunctionModelName('gpt-3.5-turbo-1106')).toBe('gpt-3.5-turbo-1106');
-      expect(client.getFunctionModelName('gpt-4-1106-preview')).toBe('gpt-4-1106-preview');
-      expect(client.getFunctionModelName('gpt-4-1106')).toBe('gpt-4-1106');
-    });
-
-    test('should return the corresponding model if input is non-function capable (`0314` models)', () => {
-      expect(client.getFunctionModelName('gpt-4-0314')).toBe('gpt-4');
-      expect(client.getFunctionModelName('gpt-4-32k-0314')).toBe('gpt-4');
-      expect(client.getFunctionModelName('gpt-3.5-turbo-0314')).toBe('gpt-3.5-turbo');
-      expect(client.getFunctionModelName('gpt-3.5-turbo-16k-0314')).toBe('gpt-3.5-turbo');
-    });
-
-    test('should return "gpt-3.5-turbo" when the input includes "gpt-3.5-turbo"', () => {
-      expect(client.getFunctionModelName('test gpt-3.5-turbo model')).toBe('gpt-3.5-turbo');
-    });
-
-    test('should return "gpt-4" when the input includes "gpt-4"', () => {
-      expect(client.getFunctionModelName('testing gpt-4')).toBe('gpt-4');
-    });
-
-    test('should return "gpt-3.5-turbo" for input that does not meet any specific condition', () => {
-      expect(client.getFunctionModelName('random string')).toBe('gpt-3.5-turbo');
-      expect(client.getFunctionModelName('')).toBe('gpt-3.5-turbo');
-    });
-  });
-
-  describe('Azure OpenAI tests specific to Plugins', () => {
-    // TODO: add more tests for Azure OpenAI integration with Plugins
-    // let client;
-    // beforeEach(() => {
-    //   client = new PluginsClient('dummy_api_key');
-    // });
-
-    test('should not call getFunctionModelName when azure options are set', () => {
-      const spy = jest.spyOn(PluginsClient.prototype, 'getFunctionModelName');
-      const model = 'gpt-4-turbo';
-
-      // note, without the azure change in PR #1766, `getFunctionModelName` is called twice
-      const testClient = new PluginsClient('dummy_api_key', {
-        agentOptions: {
-          model,
-          agent: 'functions',
-        },
-        azure: defaultAzureOptions,
-      });
-
-      expect(spy).not.toHaveBeenCalled();
-      expect(testClient.agentOptions.model).toBe(model);
-
-      spy.mockRestore();
-    });
-  });
-
-  describe('sendMessage with filtered tools', () => {
-    let TestAgent;
-    const apiKey = 'fake-api-key';
-    const mockTools = [{ name: 'tool1' }, { name: 'tool2' }, { name: 'tool3' }, { name: 'tool4' }];
-
-    beforeEach(() => {
-      TestAgent = new PluginsClient(apiKey, {
-        tools: mockTools,
-        modelOptions: {
-          model: 'gpt-3.5-turbo',
-          temperature: 0,
-          max_tokens: 2,
-        },
-        agentOptions: {
-          model: 'gpt-3.5-turbo',
-        },
-      });
-
-      TestAgent.options.req = {
-        app: {
-          locals: {},
-        },
-      };
-
-      TestAgent.sendMessage = jest.fn().mockImplementation(async () => {
-        const { filteredTools = [], includedTools = [] } = TestAgent.options.req.app.locals;
-
-        if (includedTools.length > 0) {
-          const tools = TestAgent.options.tools.filter((plugin) =>
-            includedTools.includes(plugin.name),
-          );
-          TestAgent.options.tools = tools;
-        } else {
-          const tools = TestAgent.options.tools.filter(
-            (plugin) => !filteredTools.includes(plugin.name),
-          );
-          TestAgent.options.tools = tools;
-        }
-
-        return {
-          text: 'Mocked response',
-          tools: TestAgent.options.tools,
-        };
-      });
-    });
-
-    test('should filter out tools when filteredTools is provided', async () => {
-      TestAgent.options.req.app.locals.filteredTools = ['tool1', 'tool3'];
-      const response = await TestAgent.sendMessage('Test message');
-      expect(response.tools).toHaveLength(2);
-      expect(response.tools).toEqual(
-        expect.arrayContaining([
-          expect.objectContaining({ name: 'tool2' }),
-          expect.objectContaining({ name: 'tool4' }),
-        ]),
-      );
-    });
-
-    test('should only include specified tools when includedTools is provided', async () => {
-      TestAgent.options.req.app.locals.includedTools = ['tool2', 'tool4'];
-      const response = await TestAgent.sendMessage('Test message');
-      expect(response.tools).toHaveLength(2);
-      expect(response.tools).toEqual(
-        expect.arrayContaining([
-          expect.objectContaining({ name: 'tool2' }),
-          expect.objectContaining({ name: 'tool4' }),
-        ]),
-      );
-    });
-
-    test('should prioritize includedTools over filteredTools', async () => {
-      TestAgent.options.req.app.locals.filteredTools = ['tool1', 'tool3'];
-      TestAgent.options.req.app.locals.includedTools = ['tool1', 'tool2'];
-      const response = await TestAgent.sendMessage('Test message');
-      expect(response.tools).toHaveLength(2);
-      expect(response.tools).toEqual(
-        expect.arrayContaining([
-          expect.objectContaining({ name: 'tool1' }),
-          expect.objectContaining({ name: 'tool2' }),
-        ]),
-      );
-    });
-
-    test('should not modify tools when no filters are provided', async () => {
-      const response = await TestAgent.sendMessage('Test message');
-      expect(response.tools).toHaveLength(4);
-      expect(response.tools).toEqual(expect.arrayContaining(mockTools));
-    });
-  });
-});
--- a/api/app/clients/tools/dynamic/OpenAPIPlugin.js
+++ b/api/app/clients/tools/dynamic/OpenAPIPlugin.js
@@ -1,184 +0,0 @@
-require('dotenv').config();
-const fs = require('fs');
-const { z } = require('zod');
-const path = require('path');
-const yaml = require('js-yaml');
-const { createOpenAPIChain } = require('langchain/chains');
-const { DynamicStructuredTool } = require('@langchain/core/tools');
-const { ChatPromptTemplate, HumanMessagePromptTemplate } = require('@langchain/core/prompts');
-const { logger } = require('~/config');
-
-function addLinePrefix(text, prefix = '// ') {
-  return text
-    .split('\n')
-    .map((line) => prefix + line)
-    .join('\n');
-}
-
-function createPrompt(name, functions) {
-  const prefix = `// The ${name} tool has the following functions. Determine the desired or most optimal function for the user's query:`;
-  const functionDescriptions = functions
-    .map((func) => `// - ${func.name}: ${func.description}`)
-    .join('\n');
-  return `${prefix}\n${functionDescriptions}
-// You are an expert manager and scrum master. You must provide a detailed intent to better execute the function.
-// Always format as such: {{"func": "function_name", "intent": "intent and expected result"}}`;
-}
-
-const AuthBearer = z
-  .object({
-    type: z.string().includes('service_http'),
-    authorization_type: z.string().includes('bearer'),
-    verification_tokens: z.object({
-      openai: z.string(),
-    }),
-  })
-  .catch(() => false);
-
-const AuthDefinition = z
-  .object({
-    type: z.string(),
-    authorization_type: z.string(),
-    verification_tokens: z.object({
-      openai: z.string(),
-    }),
-  })
-  .catch(() => false);
-
-async function readSpecFile(filePath) {
-  try {
-    const fileContents = await fs.promises.readFile(filePath, 'utf8');
-    if (path.extname(filePath) === '.json') {
-      return JSON.parse(fileContents);
-    }
-    return yaml.load(fileContents);
-  } catch (e) {
-    logger.error('[readSpecFile] error', e);
-    return false;
-  }
-}
-
-async function getSpec(url) {
-  const RegularUrl = z
-    .string()
-    .url()
-    .catch(() => false);
-
-  if (RegularUrl.parse(url) && path.extname(url) === '.json') {
-    const response = await fetch(url);
-    return await response.json();
-  }
-
-  const ValidSpecPath = z
-    .string()
-    .url()
-    .catch(async () => {
-      const spec = path.join(__dirname, '..', '.well-known', 'openapi', url);
-      if (!fs.existsSync(spec)) {
-        return false;
-      }
-
-      return await readSpecFile(spec);
-    });
-
-  return ValidSpecPath.parse(url);
-}
-
-async function createOpenAPIPlugin({ data, llm, user, message, memory, signal }) {
-  let spec;
-  try {
-    spec = await getSpec(data.api.url);
-  } catch (error) {
-    logger.error('[createOpenAPIPlugin] getSpec error', error);
-    return null;
-  }
-
-  if (!spec) {
-    logger.warn('[createOpenAPIPlugin] No spec found');
-    return null;
-  }
-
-  const headers = {};
-  const { auth, name_for_model, description_for_model, description_for_human } = data;
-  if (auth && AuthDefinition.parse(auth)) {
-    logger.debug('[createOpenAPIPlugin] auth detected', auth);
-    const { openai } = auth.verification_tokens;
-    if (AuthBearer.parse(auth)) {
-      headers.authorization = `Bearer ${openai}`;
-      logger.debug('[createOpenAPIPlugin] added auth bearer', headers);
-    }
-  }
-
-  const chainOptions = { llm };
-
-  if (data.headers && data.headers['librechat_user_id']) {
-    logger.debug('[createOpenAPIPlugin] id detected', headers);
-    headers[data.headers['librechat_user_id']] = user;
-  }
-
-  if (Object.keys(headers).length > 0) {
-    logger.debug('[createOpenAPIPlugin] headers detected', headers);
-    chainOptions.headers = headers;
-  }
-
-  if (data.params) {
-    logger.debug('[createOpenAPIPlugin] params detected', data.params);
-    chainOptions.params = data.params;
-  }
-
-  let history = '';
-  if (memory) {
-    logger.debug('[createOpenAPIPlugin] openAPI chain: memory detected', memory);
-    const { history: chat_history } = await memory.loadMemoryVariables({});
-    history = chat_history?.length > 0 ? `\n\n## Chat History:\n${chat_history}\n` : '';
-  }
-
-  chainOptions.prompt = ChatPromptTemplate.fromMessages([
-    HumanMessagePromptTemplate.fromTemplate(
-      `# Use the provided API's to respond to this query:\n\n{query}\n\n## Instructions:\n${addLinePrefix(
-        description_for_model,
-      )}${history}`,
-    ),
-  ]);
-
-  const chain = await createOpenAPIChain(spec, chainOptions);
-
-  const { functions } = chain.chains[0].lc_kwargs.llmKwargs;
-
-  return new DynamicStructuredTool({
-    name: name_for_model,
-    description_for_model: `${addLinePrefix(description_for_human)}${createPrompt(
-      name_for_model,
-      functions,
-    )}`,
-    description: `${description_for_human}`,
-    schema: z.object({
-      func: z
-        .string()
-        .describe(
-          `The function to invoke. The functions available are: ${functions
-            .map((func) => func.name)
-            .join(', ')}`,
-        ),
-      intent: z
-        .string()
-        .describe('Describe your intent with the function and your expected result'),
-    }),
-    func: async ({ func = '', intent = '' }) => {
-      const filteredFunctions = functions.filter((f) => f.name === func);
-      chain.chains[0].lc_kwargs.llmKwargs.functions = filteredFunctions;
-      const query = `${message}${func?.length > 0 ? `\n// Intent: ${intent}` : ''}`;
-      const result = await chain.call({
-        query,
-        signal,
-      });
-      return result.response;
-    },
-  });
-}
-
-module.exports = {
-  getSpec,
-  readSpecFile,
-  createOpenAPIPlugin,
-};
--- a/api/app/clients/tools/dynamic/OpenAPIPlugin.spec.js
+++ b/api/app/clients/tools/dynamic/OpenAPIPlugin.spec.js
@@ -1,72 +0,0 @@
-const fs = require('fs');
-const { createOpenAPIPlugin, getSpec, readSpecFile } = require('./OpenAPIPlugin');
-
-global.fetch = jest.fn().mockImplementationOnce(() => {
-  return new Promise((resolve) => {
-    resolve({
-      ok: true,
-      json: () => Promise.resolve({ key: 'value' }),
-    });
-  });
-});
-jest.mock('fs', () => ({
-  promises: {
-    readFile: jest.fn(),
-  },
-  existsSync: jest.fn(),
-}));
-
-describe('readSpecFile', () => {
-  it('reads JSON file correctly', async () => {
-    fs.promises.readFile.mockResolvedValue(JSON.stringify({ test: 'value' }));
-    const result = await readSpecFile('test.json');
-    expect(result).toEqual({ test: 'value' });
-  });
-
-  it('reads YAML file correctly', async () => {
-    fs.promises.readFile.mockResolvedValue('test: value');
-    const result = await readSpecFile('test.yaml');
-    expect(result).toEqual({ test: 'value' });
-  });
-
-  it('handles error correctly', async () => {
-    fs.promises.readFile.mockRejectedValue(new Error('test error'));
-    const result = await readSpecFile('test.json');
-    expect(result).toBe(false);
-  });
-});
-
-describe('getSpec', () => {
-  it('fetches spec from url correctly', async () => {
-    const parsedJson = await getSpec('https://www.instacart.com/.well-known/ai-plugin.json');
-    const isObject = typeof parsedJson === 'object';
-    expect(isObject).toEqual(true);
-  });
-
-  it('reads spec from file correctly', async () => {
-    fs.existsSync.mockReturnValue(true);
-    fs.promises.readFile.mockResolvedValue(JSON.stringify({ test: 'value' }));
-    const result = await getSpec('test.json');
-    expect(result).toEqual({ test: 'value' });
-  });
-
-  it('returns false when file does not exist', async () => {
-    fs.existsSync.mockReturnValue(false);
-    const result = await getSpec('test.json');
-    expect(result).toBe(false);
-  });
-});
-
-describe('createOpenAPIPlugin', () => {
-  it('returns null when getSpec throws an error', async () => {
-    const result = await createOpenAPIPlugin({ data: { api: { url: 'invalid' } } });
-    expect(result).toBe(null);
-  });
-
-  it('returns null when no spec is found', async () => {
-    const result = await createOpenAPIPlugin({});
-    expect(result).toBe(null);
-  });
-
-  // Add more tests here for different scenarios
-});
--- a/api/app/clients/tools/index.js
+++ b/api/app/clients/tools/index.js
@@ -2,13 +2,15 @@ const availableTools = require('./manifest.json');

 // Structured Tools
 const DALLE3 = require('./structured/DALLE3');
+const FluxAPI = require('./structured/FluxAPI');
 const OpenWeather = require('./structured/OpenWeather');
-const createYouTubeTools = require('./structured/YouTube');
 const StructuredWolfram = require('./structured/Wolfram');
+const createYouTubeTools = require('./structured/YouTube');
 const StructuredACS = require('./structured/AzureAISearch');
 const StructuredSD = require('./structured/StableDiffusion');
 const GoogleSearchAPI = require('./structured/GoogleSearch');
 const TraversaalSearch = require('./structured/TraversaalSearch');
+const createOpenAIImageTools = require('./structured/OpenAIImageTools');
 const TavilySearchResults = require('./structured/TavilySearchResults');

 /** @type {Record<string, TPlugin | undefined>} */
@@ -30,6 +32,7 @@ module.exports = {
  manifestToolMap,
  // Structured Tools
  DALLE3,
+  FluxAPI,
  OpenWeather,
  StructuredSD,
  StructuredACS,
@@ -38,4 +41,5 @@ module.exports = {
  StructuredWolfram,
  createYouTubeTools,
  TavilySearchResults,
+  createOpenAIImageTools,
 };
--- a/api/app/clients/tools/manifest.json
+++ b/api/app/clients/tools/manifest.json
@@ -44,6 +44,20 @@
      }
    ]
  },
+  {
+    "name": "OpenAI Image Tools",
+    "pluginKey": "image_gen_oai",
+    "toolkit": true,
+    "description": "Image Generation and Editing using OpenAI's latest state-of-the-art models",
+    "icon": "/assets/image_gen_oai.png",
+    "authConfig": [
+      {
+        "authField": "IMAGE_GEN_OAI_API_KEY",
+        "label": "OpenAI Image Tools API Key",
+        "description": "Your OpenAI API Key for Image Generation and Editing"
+      }
+    ]
+  },
  {
    "name": "Wolfram",
    "pluginKey": "wolfram",
@@ -164,5 +178,19 @@
        "description": "Sign up at <a href=\"https://home.openweathermap.org/users/sign_up\" target=\"_blank\">OpenWeather</a>, then get your key at <a href=\"https://home.openweathermap.org/api_keys\" target=\"_blank\">API keys</a>."
      }
    ]
+  },
+  {
+    "name": "Flux",
+    "pluginKey": "flux",
+    "description": "Generate images using text with the Flux API.",
+    "icon": "https://blackforestlabs.ai/wp-content/uploads/2024/07/bfl_logo_retraced_blk.png",
+    "isAuthRequired": "true",
+    "authConfig": [
+      {
+        "authField": "FLUX_API_KEY",
+        "label": "Your Flux API Key",
+        "description": "Provide your Flux API key from your user profile."
+      }
+    ]
  }
 ]
--- a/api/app/clients/tools/structured/DALLE3.js
+++ b/api/app/clients/tools/structured/DALLE3.js
@@ -1,14 +1,17 @@
 const { z } = require('zod');
 const path = require('path');
 const OpenAI = require('openai');
+const fetch = require('node-fetch');
 const { v4: uuidv4 } = require('uuid');
 const { Tool } = require('@langchain/core/tools');
 const { HttpsProxyAgent } = require('https-proxy-agent');
-const { FileContext } = require('librechat-data-provider');
+const { FileContext, ContentTypes } = require('librechat-data-provider');
 const { getImageBasename } = require('~/server/services/Files/images');
 const extractBaseURL = require('~/utils/extractBaseURL');
-const { logger } = require('~/config');
+const logger = require('~/config/winston');

+const displayMessage =
+  "DALL-E displayed an image. All generated images are already plainly visible, so don't repeat the descriptions in detail. Do not list download links as they are available in the UI already. The user may download the images by clicking on them, but do not mention anything about downloading to the user.";
 class DALLE3 extends Tool {
  constructor(fields = {}) {
    super();
@@ -114,10 +117,7 @@ class DALLE3 extends Tool {
    if (this.isAgent === true && typeof value === 'string') {
      return [value, {}];
    } else if (this.isAgent === true && typeof value === 'object') {
-      return [
-        'DALL-E displayed an image. All generated images are already plainly visible, so don\'t repeat the descriptions in detail. Do not list download links as they are available in the UI already. The user may download the images by clicking on them, but do not mention anything about downloading to the user.',
-        value,
-      ];
+      return [displayMessage, value];
    }

    return value;
@@ -160,6 +160,32 @@ Error Message: ${error.message}`);
      );
    }

+    if (this.isAgent) {
+      let fetchOptions = {};
+      if (process.env.PROXY) {
+        fetchOptions.agent = new HttpsProxyAgent(process.env.PROXY);
+      }
+      const imageResponse = await fetch(theImageUrl, fetchOptions);
+      const arrayBuffer = await imageResponse.arrayBuffer();
+      const base64 = Buffer.from(arrayBuffer).toString('base64');
+      const content = [
+        {
+          type: ContentTypes.IMAGE_URL,
+          image_url: {
+            url: `data:image/png;base64,${base64}`,
+          },
+        },
+      ];
+
+      const response = [
+        {
+          type: ContentTypes.TEXT,
+          text: displayMessage,
+        },
+      ];
+      return [response, { content }];
+    }
+
    const imageBasename = getImageBasename(theImageUrl);
    const imageExt = path.extname(imageBasename);

--- a/api/app/clients/tools/structured/FluxAPI.js
+++ b/api/app/clients/tools/structured/FluxAPI.js
@@ -0,0 +1,554 @@
+const { z } = require('zod');
+const axios = require('axios');
+const fetch = require('node-fetch');
+const { v4: uuidv4 } = require('uuid');
+const { Tool } = require('@langchain/core/tools');
+const { HttpsProxyAgent } = require('https-proxy-agent');
+const { FileContext, ContentTypes } = require('librechat-data-provider');
+const { logger } = require('~/config');
+
+const displayMessage =
+  'Flux displayed an image. All generated images are already plainly visible, so don\'t repeat the descriptions in detail. Do not list download links as they are available in the UI already. The user may download the images by clicking on them, but do not mention anything about downloading to the user.';
+
+/**
+ * FluxAPI - A tool for generating high-quality images from text prompts using the Flux API.
+ * Each call generates one image. If multiple images are needed, make multiple consecutive calls with the same or varied prompts.
+ */
+class FluxAPI extends Tool {
+  // Pricing constants in USD per image
+  static PRICING = {
+    FLUX_PRO_1_1_ULTRA: -0.06, // /v1/flux-pro-1.1-ultra
+    FLUX_PRO_1_1: -0.04, // /v1/flux-pro-1.1
+    FLUX_PRO: -0.05, // /v1/flux-pro
+    FLUX_DEV: -0.025, // /v1/flux-dev
+    FLUX_PRO_FINETUNED: -0.06, // /v1/flux-pro-finetuned
+    FLUX_PRO_1_1_ULTRA_FINETUNED: -0.07, // /v1/flux-pro-1.1-ultra-finetuned
+  };
+
+  constructor(fields = {}) {
+    super();
+
+    /** @type {boolean} Used to initialize the Tool without necessary variables. */
+    this.override = fields.override ?? false;
+
+    this.userId = fields.userId;
+    this.fileStrategy = fields.fileStrategy;
+
+    /** @type {boolean} **/
+    this.isAgent = fields.isAgent;
+    this.returnMetadata = fields.returnMetadata ?? false;
+
+    if (fields.processFileURL) {
+      /** @type {processFileURL} Necessary for output to contain all image metadata. */
+      this.processFileURL = fields.processFileURL.bind(this);
+    }
+
+    this.apiKey = fields.FLUX_API_KEY || this.getApiKey();
+
+    this.name = 'flux';
+    this.description =
+      'Use Flux to generate images from text descriptions. This tool can generate images and list available finetunes. Each generate call creates one image. For multiple images, make multiple consecutive calls.';
+
+    this.description_for_model = `// Transform any image description into a detailed, high-quality prompt. Never submit a prompt under 3 sentences. Follow these core rules:
+    // 1. ALWAYS enhance basic prompts into 5-10 detailed sentences (e.g., "a cat" becomes: "A close-up photo of a sleek Siamese cat with piercing blue eyes. The cat sits elegantly on a vintage leather armchair, its tail curled gracefully around its paws. Warm afternoon sunlight streams through a nearby window, casting gentle shadows across its face and highlighting the subtle variations in its cream and chocolate-point fur. The background is softly blurred, creating a shallow depth of field that draws attention to the cat's expressive features. The overall composition has a peaceful, contemplative mood with a professional photography style.")
+    // 2. Each prompt MUST be 3-6 descriptive sentences minimum, focusing on visual elements: lighting, composition, mood, and style
+    // Use action: 'list_finetunes' to see available custom models. When using finetunes, use endpoint: '/v1/flux-pro-finetuned' (default) or '/v1/flux-pro-1.1-ultra-finetuned' for higher quality and aspect ratio.`;
+
+    // Add base URL from environment variable with fallback
+    this.baseUrl = process.env.FLUX_API_BASE_URL || 'https://api.us1.bfl.ai';
+
+    // Define the schema for structured input
+    this.schema = z.object({
+      action: z
+        .enum(['generate', 'list_finetunes', 'generate_finetuned'])
+        .default('generate')
+        .describe(
+          'Action to perform: "generate" for image generation, "generate_finetuned" for finetuned model generation, "list_finetunes" to get available custom models',
+        ),
+      prompt: z
+        .string()
+        .optional()
+        .describe(
+          'Text prompt for image generation. Required when action is "generate". Not used for list_finetunes.',
+        ),
+      width: z
+        .number()
+        .optional()
+        .describe(
+          'Width of the generated image in pixels. Must be a multiple of 32. Default is 1024.',
+        ),
+      height: z
+        .number()
+        .optional()
+        .describe(
+          'Height of the generated image in pixels. Must be a multiple of 32. Default is 768.',
+        ),
+      prompt_upsampling: z
+        .boolean()
+        .optional()
+        .default(false)
+        .describe('Whether to perform upsampling on the prompt.'),
+      steps: z
+        .number()
+        .int()
+        .optional()
+        .describe('Number of steps to run the model for, a number from 1 to 50. Default is 40.'),
+      seed: z.number().optional().describe('Optional seed for reproducibility.'),
+      safety_tolerance: z
+        .number()
+        .optional()
+        .default(6)
+        .describe(
+          'Tolerance level for input and output moderation. Between 0 and 6, 0 being most strict, 6 being least strict.',
+        ),
+      endpoint: z
+        .enum([
+          '/v1/flux-pro-1.1',
+          '/v1/flux-pro',
+          '/v1/flux-dev',
+          '/v1/flux-pro-1.1-ultra',
+          '/v1/flux-pro-finetuned',
+          '/v1/flux-pro-1.1-ultra-finetuned',
+        ])
+        .optional()
+        .default('/v1/flux-pro-1.1')
+        .describe('Endpoint to use for image generation.'),
+      raw: z
+        .boolean()
+        .optional()
+        .default(false)
+        .describe(
+          'Generate less processed, more natural-looking images. Only works for /v1/flux-pro-1.1-ultra.',
+        ),
+      finetune_id: z.string().optional().describe('ID of the finetuned model to use'),
+      finetune_strength: z
+        .number()
+        .optional()
+        .default(1.1)
+        .describe('Strength of the finetuning effect (typically between 0.1 and 1.2)'),
+      guidance: z.number().optional().default(2.5).describe('Guidance scale for finetuned models'),
+      aspect_ratio: z
+        .string()
+        .optional()
+        .default('16:9')
+        .describe('Aspect ratio for ultra models (e.g., "16:9")'),
+    });
+  }
+
+  getAxiosConfig() {
+    const config = {};
+    if (process.env.PROXY) {
+      config.httpsAgent = new HttpsProxyAgent(process.env.PROXY);
+    }
+    return config;
+  }
+
+  /** @param {Object|string} value */
+  getDetails(value) {
+    if (typeof value === 'string') {
+      return value;
+    }
+    return JSON.stringify(value, null, 2);
+  }
+
+  getApiKey() {
+    const apiKey = process.env.FLUX_API_KEY || '';
+    if (!apiKey && !this.override) {
+      throw new Error('Missing FLUX_API_KEY environment variable.');
+    }
+    return apiKey;
+  }
+
+  wrapInMarkdown(imageUrl) {
+    const serverDomain = process.env.DOMAIN_SERVER || 'http://localhost:3080';
+    return `![generated image](${serverDomain}${imageUrl})`;
+  }
+
+  returnValue(value) {
+    if (this.isAgent === true && typeof value === 'string') {
+      return [value, {}];
+    } else if (this.isAgent === true && typeof value === 'object') {
+      if (Array.isArray(value)) {
+        return value;
+      }
+      return [displayMessage, value];
+    }
+    return value;
+  }
+
+  async _call(data) {
+    const { action = 'generate', ...imageData } = data;
+
+    // Use provided API key for this request if available, otherwise use default
+    const requestApiKey = this.apiKey || this.getApiKey();
+
+    // Handle list_finetunes action
+    if (action === 'list_finetunes') {
+      return this.getMyFinetunes(requestApiKey);
+    }
+
+    // Handle finetuned generation
+    if (action === 'generate_finetuned') {
+      return this.generateFinetunedImage(imageData, requestApiKey);
+    }
+
+    // For generate action, ensure prompt is provided
+    if (!imageData.prompt) {
+      throw new Error('Missing required field: prompt');
+    }
+
+    let payload = {
+      prompt: imageData.prompt,
+      prompt_upsampling: imageData.prompt_upsampling || false,
+      safety_tolerance: imageData.safety_tolerance || 6,
+      output_format: imageData.output_format || 'png',
+    };
+
+    // Add optional parameters if provided
+    if (imageData.width) {
+      payload.width = imageData.width;
+    }
+    if (imageData.height) {
+      payload.height = imageData.height;
+    }
+    if (imageData.steps) {
+      payload.steps = imageData.steps;
+    }
+    if (imageData.seed !== undefined) {
+      payload.seed = imageData.seed;
+    }
+    if (imageData.raw) {
+      payload.raw = imageData.raw;
+    }
+
+    const generateUrl = `${this.baseUrl}${imageData.endpoint || '/v1/flux-pro'}`;
+    const resultUrl = `${this.baseUrl}/v1/get_result`;
+
+    logger.debug('[FluxAPI] Generating image with payload:', payload);
+    logger.debug('[FluxAPI] Using endpoint:', generateUrl);
+
+    let taskResponse;
+    try {
+      taskResponse = await axios.post(generateUrl, payload, {
+        headers: {
+          'x-key': requestApiKey,
+          'Content-Type': 'application/json',
+          Accept: 'application/json',
+        },
+        ...this.getAxiosConfig(),
+      });
+    } catch (error) {
+      const details = this.getDetails(error?.response?.data || error.message);
+      logger.error('[FluxAPI] Error while submitting task:', details);
+
+      return this.returnValue(
+        `Something went wrong when trying to generate the image. The Flux API may be unavailable:
+        Error Message: ${details}`,
+      );
+    }
+
+    const taskId = taskResponse.data.id;
+
+    // Polling for the result
+    let status = 'Pending';
+    let resultData = null;
+    while (status !== 'Ready' && status !== 'Error') {
+      try {
+        // Wait 2 seconds between polls
+        await new Promise((resolve) => setTimeout(resolve, 2000));
+        const resultResponse = await axios.get(resultUrl, {
+          headers: {
+            'x-key': requestApiKey,
+            Accept: 'application/json',
+          },
+          params: { id: taskId },
+          ...this.getAxiosConfig(),
+        });
+        status = resultResponse.data.status;
+
+        if (status === 'Ready') {
+          resultData = resultResponse.data.result;
+          break;
+        } else if (status === 'Error') {
+          logger.error('[FluxAPI] Error in task:', resultResponse.data);
+          return this.returnValue('An error occurred during image generation.');
+        }
+      } catch (error) {
+        const details = this.getDetails(error?.response?.data || error.message);
+        logger.error('[FluxAPI] Error while getting result:', details);
+        return this.returnValue('An error occurred while retrieving the image.');
+      }
+    }
+
+    // If no result data
+    if (!resultData || !resultData.sample) {
+      logger.error('[FluxAPI] No image data received from API. Response:', resultData);
+      return this.returnValue('No image data received from Flux API.');
+    }
+
+    // Try saving the image locally
+    const imageUrl = resultData.sample;
+    const imageName = `img-${uuidv4()}.png`;
+
+    if (this.isAgent) {
+      try {
+        // Fetch the image and convert to base64
+        const fetchOptions = {};
+        if (process.env.PROXY) {
+          fetchOptions.agent = new HttpsProxyAgent(process.env.PROXY);
+        }
+        const imageResponse = await fetch(imageUrl, fetchOptions);
+        const arrayBuffer = await imageResponse.arrayBuffer();
+        const base64 = Buffer.from(arrayBuffer).toString('base64');
+        const content = [
+          {
+            type: ContentTypes.IMAGE_URL,
+            image_url: {
+              url: `data:image/png;base64,${base64}`,
+            },
+          },
+        ];
+
+        const response = [
+          {
+            type: ContentTypes.TEXT,
+            text: displayMessage,
+          },
+        ];
+        return [response, { content }];
+      } catch (error) {
+        logger.error('Error processing image for agent:', error);
+        return this.returnValue(`Failed to process the image. ${error.message}`);
+      }
+    }
+
+    try {
+      logger.debug('[FluxAPI] Saving image:', imageUrl);
+      const result = await this.processFileURL({
+        fileStrategy: this.fileStrategy,
+        userId: this.userId,
+        URL: imageUrl,
+        fileName: imageName,
+        basePath: 'images',
+        context: FileContext.image_generation,
+      });
+
+      logger.debug('[FluxAPI] Image saved to path:', result.filepath);
+
+      // Calculate cost based on endpoint
+      /**
+       * TODO: Cost handling
+      const endpoint = imageData.endpoint || '/v1/flux-pro';
+      const endpointKey = Object.entries(FluxAPI.PRICING).find(([key, _]) =>
+        endpoint.includes(key.toLowerCase().replace(/_/g, '-')),
+      )?.[0];
+      const cost = FluxAPI.PRICING[endpointKey] || 0;
+       */
+      this.result = this.returnMetadata ? result : this.wrapInMarkdown(result.filepath);
+      return this.returnValue(this.result);
+    } catch (error) {
+      const details = this.getDetails(error?.message ?? 'No additional error details.');
+      logger.error('Error while saving the image:', details);
+      return this.returnValue(`Failed to save the image locally. ${details}`);
+    }
+  }
+
+  async getMyFinetunes(apiKey = null) {
+    const finetunesUrl = `${this.baseUrl}/v1/my_finetunes`;
+    const detailsUrl = `${this.baseUrl}/v1/finetune_details`;
+
+    try {
+      const headers = {
+        'x-key': apiKey || this.getApiKey(),
+        'Content-Type': 'application/json',
+        Accept: 'application/json',
+      };
+
+      // Get list of finetunes
+      const response = await axios.get(finetunesUrl, {
+        headers,
+        ...this.getAxiosConfig(),
+      });
+      const finetunes = response.data.finetunes;
+
+      // Fetch details for each finetune
+      const finetuneDetails = await Promise.all(
+        finetunes.map(async (finetuneId) => {
+          try {
+            const detailResponse = await axios.get(`${detailsUrl}?finetune_id=${finetuneId}`, {
+              headers,
+              ...this.getAxiosConfig(),
+            });
+            return {
+              id: finetuneId,
+              ...detailResponse.data,
+            };
+          } catch (error) {
+            logger.error(`[FluxAPI] Error fetching details for finetune ${finetuneId}:`, error);
+            return {
+              id: finetuneId,
+              error: 'Failed to fetch details',
+            };
+          }
+        }),
+      );
+
+      if (this.isAgent) {
+        const formattedDetails = JSON.stringify(finetuneDetails, null, 2);
+        return [`Here are the available finetunes:\n${formattedDetails}`, null];
+      }
+      return JSON.stringify(finetuneDetails);
+    } catch (error) {
+      const details = this.getDetails(error?.response?.data || error.message);
+      logger.error('[FluxAPI] Error while getting finetunes:', details);
+      const errorMsg = `Failed to get finetunes: ${details}`;
+      return this.isAgent ? this.returnValue([errorMsg, {}]) : new Error(errorMsg);
+    }
+  }
+
+  async generateFinetunedImage(imageData, requestApiKey) {
+    if (!imageData.prompt) {
+      throw new Error('Missing required field: prompt');
+    }
+
+    if (!imageData.finetune_id) {
+      throw new Error(
+        'Missing required field: finetune_id for finetuned generation. Please supply a finetune_id!',
+      );
+    }
+
+    // Validate endpoint is appropriate for finetuned generation
+    const validFinetunedEndpoints = ['/v1/flux-pro-finetuned', '/v1/flux-pro-1.1-ultra-finetuned'];
+    const endpoint = imageData.endpoint || '/v1/flux-pro-finetuned';
+
+    if (!validFinetunedEndpoints.includes(endpoint)) {
+      throw new Error(
+        `Invalid endpoint for finetuned generation. Must be one of: ${validFinetunedEndpoints.join(', ')}`,
+      );
+    }
+
+    let payload = {
+      prompt: imageData.prompt,
+      prompt_upsampling: imageData.prompt_upsampling || false,
+      safety_tolerance: imageData.safety_tolerance || 6,
+      output_format: imageData.output_format || 'png',
+      finetune_id: imageData.finetune_id,
+      finetune_strength: imageData.finetune_strength || 1.0,
+      guidance: imageData.guidance || 2.5,
+    };
+
+    // Add optional parameters if provided
+    if (imageData.width) {
+      payload.width = imageData.width;
+    }
+    if (imageData.height) {
+      payload.height = imageData.height;
+    }
+    if (imageData.steps) {
+      payload.steps = imageData.steps;
+    }
+    if (imageData.seed !== undefined) {
+      payload.seed = imageData.seed;
+    }
+    if (imageData.raw) {
+      payload.raw = imageData.raw;
+    }
+
+    const generateUrl = `${this.baseUrl}${endpoint}`;
+    const resultUrl = `${this.baseUrl}/v1/get_result`;
+
+    logger.debug('[FluxAPI] Generating finetuned image with payload:', payload);
+    logger.debug('[FluxAPI] Using endpoint:', generateUrl);
+
+    let taskResponse;
+    try {
+      taskResponse = await axios.post(generateUrl, payload, {
+        headers: {
+          'x-key': requestApiKey,
+          'Content-Type': 'application/json',
+          Accept: 'application/json',
+        },
+        ...this.getAxiosConfig(),
+      });
+    } catch (error) {
+      const details = this.getDetails(error?.response?.data || error.message);
+      logger.error('[FluxAPI] Error while submitting finetuned task:', details);
+      return this.returnValue(
+        `Something went wrong when trying to generate the finetuned image. The Flux API may be unavailable:
+        Error Message: ${details}`,
+      );
+    }
+
+    const taskId = taskResponse.data.id;
+
+    // Polling for the result
+    let status = 'Pending';
+    let resultData = null;
+    while (status !== 'Ready' && status !== 'Error') {
+      try {
+        // Wait 2 seconds between polls
+        await new Promise((resolve) => setTimeout(resolve, 2000));
+        const resultResponse = await axios.get(resultUrl, {
+          headers: {
+            'x-key': requestApiKey,
+            Accept: 'application/json',
+          },
+          params: { id: taskId },
+          ...this.getAxiosConfig(),
+        });
+        status = resultResponse.data.status;
+
+        if (status === 'Ready') {
+          resultData = resultResponse.data.result;
+          break;
+        } else if (status === 'Error') {
+          logger.error('[FluxAPI] Error in finetuned task:', resultResponse.data);
+          return this.returnValue('An error occurred during finetuned image generation.');
+        }
+      } catch (error) {
+        const details = this.getDetails(error?.response?.data || error.message);
+        logger.error('[FluxAPI] Error while getting finetuned result:', details);
+        return this.returnValue('An error occurred while retrieving the finetuned image.');
+      }
+    }
+
+    // If no result data
+    if (!resultData || !resultData.sample) {
+      logger.error('[FluxAPI] No image data received from API. Response:', resultData);
+      return this.returnValue('No image data received from Flux API.');
+    }
+
+    // Try saving the image locally
+    const imageUrl = resultData.sample;
+    const imageName = `img-${uuidv4()}.png`;
+
+    try {
+      logger.debug('[FluxAPI] Saving finetuned image:', imageUrl);
+      const result = await this.processFileURL({
+        fileStrategy: this.fileStrategy,
+        userId: this.userId,
+        URL: imageUrl,
+        fileName: imageName,
+        basePath: 'images',
+        context: FileContext.image_generation,
+      });
+
+      logger.debug('[FluxAPI] Finetuned image saved to path:', result.filepath);
+
+      // Calculate cost based on endpoint
+      const endpointKey = endpoint.includes('ultra')
+        ? 'FLUX_PRO_1_1_ULTRA_FINETUNED'
+        : 'FLUX_PRO_FINETUNED';
+      const cost = FluxAPI.PRICING[endpointKey] || 0;
+      // Return the result based on returnMetadata flag
+      this.result = this.returnMetadata ? result : this.wrapInMarkdown(result.filepath);
+      return this.returnValue(this.result);
+    } catch (error) {
+      const details = this.getDetails(error?.message ?? 'No additional error details.');
+      logger.error('Error while saving the finetuned image:', details);
+      return this.returnValue(`Failed to save the finetuned image locally. ${details}`);
+    }
+  }
+}
+
+module.exports = FluxAPI;
--- a/api/app/clients/tools/structured/OpenAIImageTools.js
+++ b/api/app/clients/tools/structured/OpenAIImageTools.js
@@ -0,0 +1,548 @@
+const { z } = require('zod');
+const axios = require('axios');
+const { v4 } = require('uuid');
+const OpenAI = require('openai');
+const FormData = require('form-data');
+const { tool } = require('@langchain/core/tools');
+const { logAxiosError } = require('@librechat/api');
+const { logger } = require('@librechat/data-schemas');
+const { HttpsProxyAgent } = require('https-proxy-agent');
+const { ContentTypes, EImageOutputType } = require('librechat-data-provider');
+const { getStrategyFunctions } = require('~/server/services/Files/strategies');
+const { extractBaseURL } = require('~/utils');
+const { getFiles } = require('~/models/File');
+
+/** Default descriptions for image generation tool  */
+const DEFAULT_IMAGE_GEN_DESCRIPTION = `
+Generates high-quality, original images based solely on text, not using any uploaded reference images.
+
+When to use \`image_gen_oai\`:
+- To create entirely new images from detailed text descriptions that do NOT reference any image files.
+
+When NOT to use \`image_gen_oai\`:
+- If the user has uploaded any images and requests modifications, enhancements, or remixing based on those uploads → use \`image_edit_oai\` instead.
+
+Generated image IDs will be returned in the response, so you can refer to them in future requests made to \`image_edit_oai\`.
+`.trim();
+
+/** Default description for image editing tool  */
+const DEFAULT_IMAGE_EDIT_DESCRIPTION =
+  `Generates high-quality, original images based on text and one or more uploaded/referenced images.
+
+When to use \`image_edit_oai\`:
+- The user wants to modify, extend, or remix one **or more** uploaded images, either:
+- Previously generated, or in the current request (both to be included in the \`image_ids\` array).
+- Always when the user refers to uploaded images for editing, enhancement, remixing, style transfer, or combining elements.
+- Any current or existing images are to be used as visual guides.
+- If there are any files in the current request, they are more likely than not expected as references for image edit requests.
+
+When NOT to use \`image_edit_oai\`:
+- Brand-new generations that do not rely on an existing image → use \`image_gen_oai\` instead.
+
+Both generated and referenced image IDs will be returned in the response, so you can refer to them in future requests made to \`image_edit_oai\`.
+`.trim();
+
+/** Default prompt descriptions  */
+const DEFAULT_IMAGE_GEN_PROMPT_DESCRIPTION = `Describe the image you want in detail. 
+      Be highly specific—break your idea into layers: 
+      (1) main concept and subject,
+      (2) composition and position,
+      (3) lighting and mood,
+      (4) style, medium, or camera details,
+      (5) important features (age, expression, clothing, etc.),
+      (6) background.
+      Use positive, descriptive language and specify what should be included, not what to avoid. 
+      List number and characteristics of people/objects, and mention style/technical requirements (e.g., "DSLR photo, 85mm lens, golden hour").
+      Do not reference any uploaded images—use for new image creation from text only.`;
+
+const DEFAULT_IMAGE_EDIT_PROMPT_DESCRIPTION = `Describe the changes, enhancements, or new ideas to apply to the uploaded image(s).
+      Be highly specific—break your request into layers: 
+      (1) main concept or transformation,
+      (2) specific edits/replacements or composition guidance,
+      (3) desired style, mood, or technique,
+      (4) features/items to keep, change, or add (such as objects, people, clothing, lighting, etc.).
+      Use positive, descriptive language and clarify what should be included or changed, not what to avoid.
+      Always base this prompt on the most recently uploaded reference images.`;
+
+const displayMessage =
+  "The tool displayed an image. All generated images are already plainly visible, so don't repeat the descriptions in detail. Do not list download links as they are available in the UI already. The user may download the images by clicking on them, but do not mention anything about downloading to the user.";
+
+/**
+ * Replaces unwanted characters from the input string
+ * @param {string} inputString - The input string to process
+ * @returns {string} - The processed string
+ */
+function replaceUnwantedChars(inputString) {
+  return inputString
+    .replace(/\r\n|\r|\n/g, ' ')
+    .replace(/"/g, '')
+    .trim();
+}
+
+function returnValue(value) {
+  if (typeof value === 'string') {
+    return [value, {}];
+  } else if (typeof value === 'object') {
+    if (Array.isArray(value)) {
+      return value;
+    }
+    return [displayMessage, value];
+  }
+  return value;
+}
+
+const getImageGenDescription = () => {
+  return process.env.IMAGE_GEN_OAI_DESCRIPTION || DEFAULT_IMAGE_GEN_DESCRIPTION;
+};
+
+const getImageEditDescription = () => {
+  return process.env.IMAGE_EDIT_OAI_DESCRIPTION || DEFAULT_IMAGE_EDIT_DESCRIPTION;
+};
+
+const getImageGenPromptDescription = () => {
+  return process.env.IMAGE_GEN_OAI_PROMPT_DESCRIPTION || DEFAULT_IMAGE_GEN_PROMPT_DESCRIPTION;
+};
+
+const getImageEditPromptDescription = () => {
+  return process.env.IMAGE_EDIT_OAI_PROMPT_DESCRIPTION || DEFAULT_IMAGE_EDIT_PROMPT_DESCRIPTION;
+};
+
+function createAbortHandler() {
+  return function () {
+    logger.debug('[ImageGenOAI] Image generation aborted');
+  };
+}
+
+/**
+ * Creates OpenAI Image tools (generation and editing)
+ * @param {Object} fields - Configuration fields
+ * @param {ServerRequest} fields.req - Whether the tool is being used in an agent context
+ * @param {boolean} fields.isAgent - Whether the tool is being used in an agent context
+ * @param {string} fields.IMAGE_GEN_OAI_API_KEY - The OpenAI API key
+ * @param {boolean} [fields.override] - Whether to override the API key check, necessary for app initialization
+ * @param {MongoFile[]} [fields.imageFiles] - The images to be used for editing
+ * @returns {Array} - Array of image tools
+ */
+function createOpenAIImageTools(fields = {}) {
+  /** @type {boolean} Used to initialize the Tool without necessary variables. */
+  const override = fields.override ?? false;
+  /** @type {boolean} */
+  if (!override && !fields.isAgent) {
+    throw new Error('This tool is only available for agents.');
+  }
+  const { req } = fields;
+  const imageOutputType = req?.app.locals.imageOutputType || EImageOutputType.PNG;
+  const appFileStrategy = req?.app.locals.fileStrategy;
+
+  const getApiKey = () => {
+    const apiKey = process.env.IMAGE_GEN_OAI_API_KEY ?? '';
+    if (!apiKey && !override) {
+      throw new Error('Missing IMAGE_GEN_OAI_API_KEY environment variable.');
+    }
+    return apiKey;
+  };
+
+  let apiKey = fields.IMAGE_GEN_OAI_API_KEY ?? getApiKey();
+  const closureConfig = { apiKey };
+
+  let baseURL = 'https://api.openai.com/v1/';
+  if (!override && process.env.IMAGE_GEN_OAI_BASEURL) {
+    baseURL = extractBaseURL(process.env.IMAGE_GEN_OAI_BASEURL);
+    closureConfig.baseURL = baseURL;
+  }
+
+  // Note: Azure may not yet support the latest image generation models
+  if (
+    !override &&
+    process.env.IMAGE_GEN_OAI_AZURE_API_VERSION &&
+    process.env.IMAGE_GEN_OAI_BASEURL
+  ) {
+    baseURL = process.env.IMAGE_GEN_OAI_BASEURL;
+    closureConfig.baseURL = baseURL;
+    closureConfig.defaultQuery = { 'api-version': process.env.IMAGE_GEN_OAI_AZURE_API_VERSION };
+    closureConfig.defaultHeaders = {
+      'api-key': process.env.IMAGE_GEN_OAI_API_KEY,
+      'Content-Type': 'application/json',
+    };
+    closureConfig.apiKey = process.env.IMAGE_GEN_OAI_API_KEY;
+  }
+
+  const imageFiles = fields.imageFiles ?? [];
+
+  /**
+   * Image Generation Tool
+   */
+  const imageGenTool = tool(
+    async (
+      {
+        prompt,
+        background = 'auto',
+        n = 1,
+        output_compression = 100,
+        quality = 'auto',
+        size = 'auto',
+      },
+      runnableConfig,
+    ) => {
+      if (!prompt) {
+        throw new Error('Missing required field: prompt');
+      }
+      const clientConfig = { ...closureConfig };
+      if (process.env.PROXY) {
+        clientConfig.httpAgent = new HttpsProxyAgent(process.env.PROXY);
+      }
+
+      /** @type {OpenAI} */
+      const openai = new OpenAI(clientConfig);
+      let output_format = imageOutputType;
+      if (
+        background === 'transparent' &&
+        output_format !== EImageOutputType.PNG &&
+        output_format !== EImageOutputType.WEBP
+      ) {
+        logger.warn(
+          '[ImageGenOAI] Transparent background requires PNG or WebP format, defaulting to PNG',
+        );
+        output_format = EImageOutputType.PNG;
+      }
+
+      let resp;
+      /** @type {AbortSignal} */
+      let derivedSignal = null;
+      /** @type {() => void} */
+      let abortHandler = null;
+
+      try {
+        if (runnableConfig?.signal) {
+          derivedSignal = AbortSignal.any([runnableConfig.signal]);
+          abortHandler = createAbortHandler();
+          derivedSignal.addEventListener('abort', abortHandler, { once: true });
+        }
+
+        resp = await openai.images.generate(
+          {
+            model: 'gpt-image-1',
+            prompt: replaceUnwantedChars(prompt),
+            n: Math.min(Math.max(1, n), 10),
+            background,
+            output_format,
+            output_compression:
+              output_format === EImageOutputType.WEBP || output_format === EImageOutputType.JPEG
+                ? output_compression
+                : undefined,
+            quality,
+            size,
+          },
+          {
+            signal: derivedSignal,
+          },
+        );
+      } catch (error) {
+        const message = '[image_gen_oai] Problem generating the image:';
+        logAxiosError({ error, message });
+        return returnValue(`Something went wrong when trying to generate the image. The OpenAI API may be unavailable:
+Error Message: ${error.message}`);
+      } finally {
+        if (abortHandler && derivedSignal) {
+          derivedSignal.removeEventListener('abort', abortHandler);
+        }
+      }
+
+      if (!resp) {
+        return returnValue(
+          'Something went wrong when trying to generate the image. The OpenAI API may be unavailable',
+        );
+      }
+
+      // For gpt-image-1, the response contains base64-encoded images
+      // TODO: handle cost in `resp.usage`
+      const base64Image = resp.data[0].b64_json;
+
+      if (!base64Image) {
+        return returnValue(
+          'No image data returned from OpenAI API. There may be a problem with the API or your configuration.',
+        );
+      }
+
+      const content = [
+        {
+          type: ContentTypes.IMAGE_URL,
+          image_url: {
+            url: `data:image/${output_format};base64,${base64Image}`,
+          },
+        },
+      ];
+
+      const file_ids = [v4()];
+      const response = [
+        {
+          type: ContentTypes.TEXT,
+          text: displayMessage + `\n\ngenerated_image_id: "${file_ids[0]}"`,
+        },
+      ];
+      return [response, { content, file_ids }];
+    },
+    {
+      name: 'image_gen_oai',
+      description: getImageGenDescription(),
+      schema: z.object({
+        prompt: z.string().max(32000).describe(getImageGenPromptDescription()),
+        background: z
+          .enum(['transparent', 'opaque', 'auto'])
+          .optional()
+          .describe(
+            'Sets transparency for the background. Must be one of transparent, opaque or auto (default). When transparent, the output format should be png or webp.',
+          ),
+        /*
+        n: z
+          .number()
+          .int()
+          .min(1)
+          .max(10)
+          .optional()
+          .describe('The number of images to generate. Must be between 1 and 10.'),
+        output_compression: z
+          .number()
+          .int()
+          .min(0)
+          .max(100)
+          .optional()
+          .describe('The compression level (0-100%) for webp or jpeg formats. Defaults to 100.'),
+           */
+        quality: z
+          .enum(['auto', 'high', 'medium', 'low'])
+          .optional()
+          .describe('The quality of the image. One of auto (default), high, medium, or low.'),
+        size: z
+          .enum(['auto', '1024x1024', '1536x1024', '1024x1536'])
+          .optional()
+          .describe(
+            'The size of the generated image. One of 1024x1024, 1536x1024 (landscape), 1024x1536 (portrait), or auto (default).',
+          ),
+      }),
+      responseFormat: 'content_and_artifact',
+    },
+  );
+
+  /**
+   * Image Editing Tool
+   */
+  const imageEditTool = tool(
+    async ({ prompt, image_ids, quality = 'auto', size = 'auto' }, runnableConfig) => {
+      if (!prompt) {
+        throw new Error('Missing required field: prompt');
+      }
+
+      const clientConfig = { ...closureConfig };
+      if (process.env.PROXY) {
+        clientConfig.httpAgent = new HttpsProxyAgent(process.env.PROXY);
+      }
+
+      const formData = new FormData();
+      formData.append('model', 'gpt-image-1');
+      formData.append('prompt', replaceUnwantedChars(prompt));
+      // TODO: `mask` support
+      // TODO: more than 1 image support
+      // formData.append('n', n.toString());
+      formData.append('quality', quality);
+      formData.append('size', size);
+
+      /** @type {Record<FileSources, undefined | NodeStreamDownloader<File>>} */
+      const streamMethods = {};
+
+      const requestFilesMap = Object.fromEntries(imageFiles.map((f) => [f.file_id, { ...f }]));
+
+      const orderedFiles = new Array(image_ids.length);
+      const idsToFetch = [];
+      const indexOfMissing = Object.create(null);
+
+      for (let i = 0; i < image_ids.length; i++) {
+        const id = image_ids[i];
+        const file = requestFilesMap[id];
+
+        if (file) {
+          orderedFiles[i] = file;
+        } else {
+          idsToFetch.push(id);
+          indexOfMissing[id] = i;
+        }
+      }
+
+      if (idsToFetch.length) {
+        const fetchedFiles = await getFiles(
+          {
+            user: req.user.id,
+            file_id: { $in: idsToFetch },
+            height: { $exists: true },
+            width: { $exists: true },
+          },
+          {},
+          {},
+        );
+
+        for (const file of fetchedFiles) {
+          requestFilesMap[file.file_id] = file;
+          orderedFiles[indexOfMissing[file.file_id]] = file;
+        }
+      }
+      for (const imageFile of orderedFiles) {
+        if (!imageFile) {
+          continue;
+        }
+        /** @type {NodeStream<File>} */
+        let stream;
+        /** @type {NodeStreamDownloader<File>} */
+        let getDownloadStream;
+        const source = imageFile.source || appFileStrategy;
+        if (!source) {
+          throw new Error('No source found for image file');
+        }
+        if (streamMethods[source]) {
+          getDownloadStream = streamMethods[source];
+        } else {
+          ({ getDownloadStream } = getStrategyFunctions(source));
+          streamMethods[source] = getDownloadStream;
+        }
+        if (!getDownloadStream) {
+          throw new Error(`No download stream method found for source: ${source}`);
+        }
+        stream = await getDownloadStream(req, imageFile.filepath);
+        if (!stream) {
+          throw new Error('Failed to get download stream for image file');
+        }
+        formData.append('image[]', stream, {
+          filename: imageFile.filename,
+          contentType: imageFile.type,
+        });
+      }
+
+      /** @type {import('axios').RawAxiosHeaders} */
+      let headers = {
+        ...formData.getHeaders(),
+      };
+
+      if (process.env.IMAGE_GEN_OAI_AZURE_API_VERSION && process.env.IMAGE_GEN_OAI_BASEURL) {
+        headers['api-key'] = apiKey;
+      } else {
+        headers['Authorization'] = `Bearer ${apiKey}`;
+      }
+
+      /** @type {AbortSignal} */
+      let derivedSignal = null;
+      /** @type {() => void} */
+      let abortHandler = null;
+
+      try {
+        if (runnableConfig?.signal) {
+          derivedSignal = AbortSignal.any([runnableConfig.signal]);
+          abortHandler = createAbortHandler();
+          derivedSignal.addEventListener('abort', abortHandler, { once: true });
+        }
+
+        /** @type {import('axios').AxiosRequestConfig} */
+        const axiosConfig = {
+          headers,
+          ...clientConfig,
+          signal: derivedSignal,
+          baseURL,
+        };
+
+        if (process.env.IMAGE_GEN_OAI_AZURE_API_VERSION && process.env.IMAGE_GEN_OAI_BASEURL) {
+          axiosConfig.params = {
+            'api-version': process.env.IMAGE_GEN_OAI_AZURE_API_VERSION,
+            ...axiosConfig.params,
+          };
+        }
+        const response = await axios.post('/images/edits', formData, axiosConfig);
+
+        if (!response.data || !response.data.data || !response.data.data.length) {
+          return returnValue(
+            'No image data returned from OpenAI API. There may be a problem with the API or your configuration.',
+          );
+        }
+
+        const base64Image = response.data.data[0].b64_json;
+        if (!base64Image) {
+          return returnValue(
+            'No image data returned from OpenAI API. There may be a problem with the API or your configuration.',
+          );
+        }
+
+        const content = [
+          {
+            type: ContentTypes.IMAGE_URL,
+            image_url: {
+              url: `data:image/${imageOutputType};base64,${base64Image}`,
+            },
+          },
+        ];
+
+        const file_ids = [v4()];
+        const textResponse = [
+          {
+            type: ContentTypes.TEXT,
+            text:
+              displayMessage +
+              `\n\ngenerated_image_id: "${file_ids[0]}"\nreferenced_image_ids: ["${image_ids.join('", "')}"]`,
+          },
+        ];
+        return [textResponse, { content, file_ids }];
+      } catch (error) {
+        const message = '[image_edit_oai] Problem editing the image:';
+        logAxiosError({ error, message });
+        return returnValue(`Something went wrong when trying to edit the image. The OpenAI API may be unavailable:
+Error Message: ${error.message || 'Unknown error'}`);
+      } finally {
+        if (abortHandler && derivedSignal) {
+          derivedSignal.removeEventListener('abort', abortHandler);
+        }
+      }
+    },
+    {
+      name: 'image_edit_oai',
+      description: getImageEditDescription(),
+      schema: z.object({
+        image_ids: z
+          .array(z.string())
+          .min(1)
+          .describe(
+            `
+IDs (image ID strings) of previously generated or uploaded images that should guide the edit.
+
+Guidelines:
+- If the user's request depends on any prior image(s), copy their image IDs into the \`image_ids\` array (in the same order the user refers to them).  
+- Never invent or hallucinate IDs; only use IDs that are still visible in the conversation context.
+- If no earlier image is relevant, omit the field entirely.
+`.trim(),
+          ),
+        prompt: z.string().max(32000).describe(getImageEditPromptDescription()),
+        /*
+        n: z
+          .number()
+          .int()
+          .min(1)
+          .max(10)
+          .optional()
+          .describe('The number of images to generate. Must be between 1 and 10. Defaults to 1.'),
+        */
+        quality: z
+          .enum(['auto', 'high', 'medium', 'low'])
+          .optional()
+          .describe(
+            'The quality of the image. One of auto (default), high, medium, or low. High/medium/low only supported for gpt-image-1.',
+          ),
+        size: z
+          .enum(['auto', '1024x1024', '1536x1024', '1024x1536', '256x256', '512x512'])
+          .optional()
+          .describe(
+            'The size of the generated images. For gpt-image-1: auto (default), 1024x1024, 1536x1024, 1024x1536. For dall-e-2: 256x256, 512x512, 1024x1024.',
+          ),
+      }),
+      responseFormat: 'content_and_artifact',
+    },
+  );
+
+  return [imageGenTool, imageEditTool];
+}
+
+module.exports = createOpenAIImageTools;
--- a/api/app/clients/tools/structured/StableDiffusion.js
+++ b/api/app/clients/tools/structured/StableDiffusion.js
@@ -6,10 +6,13 @@ const axios = require('axios');
 const sharp = require('sharp');
 const { v4: uuidv4 } = require('uuid');
 const { Tool } = require('@langchain/core/tools');
-const { FileContext } = require('librechat-data-provider');
+const { FileContext, ContentTypes } = require('librechat-data-provider');
 const paths = require('~/config/paths');
 const { logger } = require('~/config');

+const displayMessage =
+  'Stable Diffusion displayed an image. All generated images are already plainly visible, so don\'t repeat the descriptions in detail. Do not list download links as they are available in the UI already. The user may download the images by clicking on them, but do not mention anything about downloading to the user.';
+
 class StableDiffusionAPI extends Tool {
  constructor(fields) {
    super();
@@ -21,6 +24,8 @@ class StableDiffusionAPI extends Tool {
    this.override = fields.override ?? false;
    /** @type {boolean} Necessary for output to contain all image metadata. */
    this.returnMetadata = fields.returnMetadata ?? false;
+    /** @type {boolean} */
+    this.isAgent = fields.isAgent;
    if (fields.uploadImageBuffer) {
      /** @type {uploadImageBuffer} Necessary for output to contain all image metadata. */
      this.uploadImageBuffer = fields.uploadImageBuffer.bind(this);
@@ -66,6 +71,16 @@ class StableDiffusionAPI extends Tool {
    return `![generated image](/${imageUrl})`;
  }

+  returnValue(value) {
+    if (this.isAgent === true && typeof value === 'string') {
+      return [value, {}];
+    } else if (this.isAgent === true && typeof value === 'object') {
+      return [displayMessage, value];
+    }
+
+    return value;
+  }
+
  getServerURL() {
    const url = process.env.SD_WEBUI_URL || '';
    if (!url && !this.override) {
@@ -113,6 +128,25 @@ class StableDiffusionAPI extends Tool {
    }

    try {
+      if (this.isAgent) {
+        const content = [
+          {
+            type: ContentTypes.IMAGE_URL,
+            image_url: {
+              url: `data:image/png;base64,${image}`,
+            },
+          },
+        ];
+
+        const response = [
+          {
+            type: ContentTypes.TEXT,
+            text: displayMessage,
+          },
+        ];
+        return [response, { content }];
+      }
+
      const buffer = Buffer.from(image.split(',', 1)[0], 'base64');
      if (this.returnMetadata && this.uploadImageBuffer && this.req) {
        const file = await this.uploadImageBuffer({
@@ -154,7 +188,7 @@ class StableDiffusionAPI extends Tool {
      logger.error('[StableDiffusion] Error while saving the image:', error);
    }

-    return this.result;
+    return this.returnValue(this.result);
  }
 }

--- a/api/app/clients/tools/structured/TavilySearchResults.js
+++ b/api/app/clients/tools/structured/TavilySearchResults.js
@@ -43,9 +43,39 @@ class TavilySearchResults extends Tool {
        .boolean()
        .optional()
        .describe('Whether to include answers in the search results. Default is False.'),
-      // include_raw_content: z.boolean().optional().describe('Whether to include raw content in the search results. Default is False.'),
-      // include_domains: z.array(z.string()).optional().describe('A list of domains to specifically include in the search results.'),
-      // exclude_domains: z.array(z.string()).optional().describe('A list of domains to specifically exclude from the search results.'),
+      include_raw_content: z
+        .boolean()
+        .optional()
+        .describe('Whether to include raw content in the search results. Default is False.'),
+      include_domains: z
+        .array(z.string())
+        .optional()
+        .describe('A list of domains to specifically include in the search results.'),
+      exclude_domains: z
+        .array(z.string())
+        .optional()
+        .describe('A list of domains to specifically exclude from the search results.'),
+      topic: z
+        .enum(['general', 'news', 'finance'])
+        .optional()
+        .describe(
+          'The category of the search. Use news ONLY if query SPECIFCALLY mentions the word "news".',
+        ),
+      time_range: z
+        .enum(['day', 'week', 'month', 'year', 'd', 'w', 'm', 'y'])
+        .optional()
+        .describe('The time range back from the current date to filter results.'),
+      days: z
+        .number()
+        .min(1)
+        .optional()
+        .describe('Number of days back from the current date to include. Only if topic is news.'),
+      include_image_descriptions: z
+        .boolean()
+        .optional()
+        .describe(
+          'When include_images is true, also add a descriptive text for each image. Default is false.',
+        ),
    });
  }

--- a/api/app/clients/tools/structured/specs/DALLE3.spec.js
+++ b/api/app/clients/tools/structured/specs/DALLE3.spec.js
@@ -1,10 +1,29 @@
 const OpenAI = require('openai');
 const DALLE3 = require('../DALLE3');
-
-const { logger } = require('~/config');
+const logger = require('~/config/winston');

 jest.mock('openai');

+jest.mock('@librechat/data-schemas', () => {
+  return {
+    logger: {
+      info: jest.fn(),
+      warn: jest.fn(),
+      debug: jest.fn(),
+      error: jest.fn(),
+    },
+  };
+});
+
+jest.mock('tiktoken', () => {
+  return {
+    encoding_for_model: jest.fn().mockReturnValue({
+      encode: jest.fn(),
+      decode: jest.fn(),
+    }),
+  };
+});
+
 const processFileURL = jest.fn();

 jest.mock('~/server/services/Files/images', () => ({
@@ -37,6 +56,11 @@ jest.mock('fs', () => {
  return {
    existsSync: jest.fn(),
    mkdirSync: jest.fn(),
+    promises: {
+      writeFile: jest.fn(),
+      readFile: jest.fn(),
+      unlink: jest.fn(),
+    },
  };
 });

--- a/api/app/clients/tools/util/addOpenAPISpecs.js
+++ b/api/app/clients/tools/util/addOpenAPISpecs.js
@@ -1,30 +0,0 @@
-const { loadSpecs } = require('./loadSpecs');
-
-function transformSpec(input) {
-  return {
-    name: input.name_for_human,
-    pluginKey: input.name_for_model,
-    description: input.description_for_human,
-    icon: input?.logo_url ?? 'https://placehold.co/70x70.png',
-    // TODO: add support for authentication
-    isAuthRequired: 'false',
-    authConfig: [],
-  };
-}
-
-async function addOpenAPISpecs(availableTools) {
-  try {
-    const specs = (await loadSpecs({})).map(transformSpec);
-    if (specs.length > 0) {
-      return [...specs, ...availableTools];
-    }
-    return availableTools;
-  } catch (error) {
-    return availableTools;
-  }
-}
-
-module.exports = {
-  transformSpec,
-  addOpenAPISpecs,
-};
--- a/api/app/clients/tools/util/addOpenAPISpecs.spec.js
+++ b/api/app/clients/tools/util/addOpenAPISpecs.spec.js
@@ -1,76 +0,0 @@
-const { addOpenAPISpecs, transformSpec } = require('./addOpenAPISpecs');
-const { loadSpecs } = require('./loadSpecs');
-const { createOpenAPIPlugin } = require('../dynamic/OpenAPIPlugin');
-
-jest.mock('./loadSpecs');
-jest.mock('../dynamic/OpenAPIPlugin');
-
-describe('transformSpec', () => {
-  it('should transform input spec to a desired format', () => {
-    const input = {
-      name_for_human: 'Human Name',
-      name_for_model: 'Model Name',
-      description_for_human: 'Human Description',
-      logo_url: 'https://example.com/logo.png',
-    };
-
-    const expectedOutput = {
-      name: 'Human Name',
-      pluginKey: 'Model Name',
-      description: 'Human Description',
-      icon: 'https://example.com/logo.png',
-      isAuthRequired: 'false',
-      authConfig: [],
-    };
-
-    expect(transformSpec(input)).toEqual(expectedOutput);
-  });
-
-  it('should use default icon if logo_url is not provided', () => {
-    const input = {
-      name_for_human: 'Human Name',
-      name_for_model: 'Model Name',
-      description_for_human: 'Human Description',
-    };
-
-    const expectedOutput = {
-      name: 'Human Name',
-      pluginKey: 'Model Name',
-      description: 'Human Description',
-      icon: 'https://placehold.co/70x70.png',
-      isAuthRequired: 'false',
-      authConfig: [],
-    };
-
-    expect(transformSpec(input)).toEqual(expectedOutput);
-  });
-});
-
-describe('addOpenAPISpecs', () => {
-  it('should add specs to available tools', async () => {
-    const availableTools = ['Tool1', 'Tool2'];
-    const specs = [
-      {
-        name_for_human: 'Human Name',
-        name_for_model: 'Model Name',
-        description_for_human: 'Human Description',
-        logo_url: 'https://example.com/logo.png',
-      },
-    ];
-
-    loadSpecs.mockResolvedValue(specs);
-    createOpenAPIPlugin.mockReturnValue('Plugin');
-
-    const result = await addOpenAPISpecs(availableTools);
-    expect(result).toEqual([...specs.map(transformSpec), ...availableTools]);
-  });
-
-  it('should return available tools if specs loading fails', async () => {
-    const availableTools = ['Tool1', 'Tool2'];
-
-    loadSpecs.mockRejectedValue(new Error('Failed to load specs'));
-
-    const result = await addOpenAPISpecs(availableTools);
-    expect(result).toEqual(availableTools);
-  });
-});
--- a/api/app/clients/tools/util/fileSearch.js
+++ b/api/app/clients/tools/util/fileSearch.js
@@ -1,26 +1,35 @@
 const { z } = require('zod');
 const axios = require('axios');
 const { tool } = require('@langchain/core/tools');
+const { logger } = require('@librechat/data-schemas');
 const { Tools, EToolResources } = require('librechat-data-provider');
+const { generateShortLivedToken } = require('~/server/services/AuthService');
 const { getFiles } = require('~/models/File');
-const { logger } = require('~/config');

 /**
 *
 * @param {Object} options
 * @param {ServerRequest} options.req
 * @param {Agent['tool_resources']} options.tool_resources
+ * @param {string} [options.agentId] - The agent ID for file access control
 * @returns {Promise<{
 *   files: Array<{ file_id: string; filename: string }>,
 *   toolContext: string
 * }>}
 */
 const primeFiles = async (options) => {
-  const { tool_resources } = options;
+  const { tool_resources, req, agentId } = options;
  const file_ids = tool_resources?.[EToolResources.file_search]?.file_ids ?? [];
  const agentResourceIds = new Set(file_ids);
  const resourceFiles = tool_resources?.[EToolResources.file_search]?.files ?? [];
-  const dbFiles = ((await getFiles({ file_id: { $in: file_ids } })) ?? []).concat(resourceFiles);
+  const dbFiles = (
+    (await getFiles(
+      { file_id: { $in: file_ids } },
+      null,
+      { text: 0 },
+      { userId: req?.user?.id, agentId },
+    )) ?? []
+  ).concat(resourceFiles);

  let toolContext = `- Note: Semantic search is available through the ${Tools.file_search} tool but no files are currently loaded. Request the user to upload documents to search through.`;

@@ -59,7 +68,7 @@ const createFileSearchTool = async ({ req, files, entity_id }) => {
      if (files.length === 0) {
        return 'No files to search. Instruct the user to add files for the search.';
      }
-      const jwtToken = req.headers.authorization.split(' ')[1];
+      const jwtToken = generateShortLivedToken(req.user.id);
      if (!jwtToken) {
        return 'There was an error authenticating the file search request.';
      }
@@ -135,7 +144,7 @@ const createFileSearchTool = async ({ req, files, entity_id }) => {
        query: z
          .string()
          .describe(
-            'A natural language query to search for relevant information in the files. Be specific and use keywords related to the information you\'re looking for. The query will be used for semantic similarity matching against the file contents.',
+            "A natural language query to search for relevant information in the files. Be specific and use keywords related to the information you're looking for. The query will be used for semantic similarity matching against the file contents.",
          ),
      }),
    },
--- a/api/app/clients/tools/util/handleTools.js
+++ b/api/app/clients/tools/util/handleTools.js
@@ -1,8 +1,9 @@
-const { Tools, Constants } = require('librechat-data-provider');
+const { logger } = require('@librechat/data-schemas');
 const { SerpAPI } = require('@langchain/community/tools/serpapi');
 const { Calculator } = require('@langchain/community/tools/calculator');
-const { createCodeExecutionTool, EnvVar } = require('@librechat/agents');
-const { getUserPluginAuthValue } = require('~/server/services/PluginService');
+const { mcpToolPattern, loadWebSearchAuth } = require('@librechat/api');
+const { EnvVar, createCodeExecutionTool, createSearchTool } = require('@librechat/agents');
+const { Tools, EToolResources, replaceSpecialVars } = require('librechat-data-provider');
 const {
  availableTools,
  manifestToolMap,
@@ -10,6 +11,7 @@ const {
  GoogleSearchAPI,
  // Structured Tools
  DALLE3,
+  FluxAPI,
  OpenWeather,
  StructuredSD,
  StructuredACS,
@@ -17,14 +19,14 @@ const {
  StructuredWolfram,
  createYouTubeTools,
  TavilySearchResults,
+  createOpenAIImageTools,
 } = require('../');
 const { primeFiles: primeCodeFiles } = require('~/server/services/Files/Code/process');
 const { createFileSearchTool, primeFiles: primeSearchFiles } = require('./fileSearch');
+const { getUserPluginAuthValue } = require('~/server/services/PluginService');
+const { loadAuthValues } = require('~/server/services/Tools/credentials');
+const { getCachedTools } = require('~/server/services/Config');
 const { createMCPTool } = require('~/server/services/MCP');
-const { loadSpecs } = require('./loadSpecs');
-const { logger } = require('~/config');
-
-const mcpToolPattern = new RegExp(`^.+${Constants.mcp_delimiter}.+$`);

 /**
 * Validates the availability and authentication of tools for a user based on environment variables or user-specific plugin authentication values.
@@ -85,49 +87,10 @@ const validateTools = async (user, tools = []) => {
    return Array.from(validToolsSet.values());
  } catch (err) {
    logger.error('[validateTools] There was a problem validating tools', err);
-    throw new Error('There was a problem validating tools');
+    throw new Error(err);
  }
 };

-const loadAuthValues = async ({ userId, authFields, throwError = true }) => {
-  let authValues = {};
-
-  /**
-   * Finds the first non-empty value for the given authentication field, supporting alternate fields.
-   * @param {string[]} fields Array of strings representing the authentication fields. Supports alternate fields delimited by "||".
-   * @returns {Promise<{ authField: string, authValue: string} | null>} An object containing the authentication field and value, or null if not found.
-   */
-  const findAuthValue = async (fields) => {
-    for (const field of fields) {
-      let value = process.env[field];
-      if (value) {
-        return { authField: field, authValue: value };
-      }
-      try {
-        value = await getUserPluginAuthValue(userId, field, throwError);
-      } catch (err) {
-        if (field === fields[fields.length - 1] && !value) {
-          throw err;
-        }
-      }
-      if (value) {
-        return { authField: field, authValue: value };
-      }
-    }
-    return null;
-  };
-
-  for (let authField of authFields) {
-    const fields = authField.split('||');
-    const result = await findAuthValue(fields);
-    if (result) {
-      authValues[result.authField] = result.authValue;
-    }
-  }
-
-  return authValues;
-};
-
 /** @typedef {typeof import('@langchain/core/tools').Tool} ToolConstructor */
 /** @typedef {import('@langchain/core/tools').Tool} Tool */

@@ -160,7 +123,7 @@ const getAuthFields = (toolKey) => {
 *
 * @param {object} object
 * @param {string} object.user
- * @param {Agent} [object.agent]
+ * @param {Pick<Agent, 'id' | 'provider' | 'model'>} [object.agent]
 * @param {string} [object.model]
 * @param {EModelEndpoint} [object.endpoint]
 * @param {LoadToolOptions} [object.options]
@@ -175,13 +138,13 @@ const loadTools = async ({
  agent,
  model,
  endpoint,
-  useSpecs,
  tools = [],
  options = {},
  functions = true,
  returnMap = false,
 }) => {
  const toolConstructors = {
+    flux: FluxAPI,
    calculator: Calculator,
    google: GoogleSearchAPI,
    open_weather: OpenWeather,
@@ -193,7 +156,7 @@ const loadTools = async ({
  };

  const customConstructors = {
-    serpapi: async () => {
+    serpapi: async (_toolContextMap) => {
      const authFields = getAuthFields('serpapi');
      let envVar = authFields[0] ?? '';
      let apiKey = process.env[envVar];
@@ -206,11 +169,40 @@ const loadTools = async ({
        gl: 'us',
      });
    },
-    youtube: async () => {
+    youtube: async (_toolContextMap) => {
      const authFields = getAuthFields('youtube');
      const authValues = await loadAuthValues({ userId: user, authFields });
      return createYouTubeTools(authValues);
    },
+    image_gen_oai: async (toolContextMap) => {
+      const authFields = getAuthFields('image_gen_oai');
+      const authValues = await loadAuthValues({ userId: user, authFields });
+      const imageFiles = options.tool_resources?.[EToolResources.image_edit]?.files ?? [];
+      let toolContext = '';
+      for (let i = 0; i < imageFiles.length; i++) {
+        const file = imageFiles[i];
+        if (!file) {
+          continue;
+        }
+        if (i === 0) {
+          toolContext =
+            'Image files provided in this request (their image IDs listed in order of appearance) available for image editing:';
+        }
+        toolContext += `\n\t- ${file.file_id}`;
+        if (i === imageFiles.length - 1) {
+          toolContext += `\n\nInclude any you need in the \`image_ids\` array when calling \`${EToolResources.image_edit}_oai\`. You may also include previously referenced or generated image IDs.`;
+        }
+      }
+      if (toolContext) {
+        toolContextMap.image_edit_oai = toolContext;
+      }
+      return createOpenAIImageTools({
+        ...authValues,
+        isAgent: !!agent,
+        req: options.req,
+        imageFiles,
+      });
+    },
  };

  const requestedTools = {};
@@ -230,14 +222,15 @@ const loadTools = async ({
  };

  const toolOptions = {
-    serpapi: { location: 'Austin,Texas,United States', hl: 'en', gl: 'us' },
+    flux: imageGenOptions,
    dalle: imageGenOptions,
    'stable-diffusion': imageGenOptions,
+    serpapi: { location: 'Austin,Texas,United States', hl: 'en', gl: 'us' },
  };

+  /** @type {Record<string, string>} */
  const toolContextMap = {};
-  const remainingTools = [];
-  const appTools = options.req?.app?.locals?.availableTools ?? {};
+  const cachedTools = (await getCachedTools({ userId: user, includeGlobal: true })) ?? {};

  for (const tool of tools) {
    if (tool === Tools.execute_code) {
@@ -247,7 +240,13 @@ const loadTools = async ({
          authFields: [EnvVar.CODE_API_KEY],
        });
        const codeApiKey = authValues[EnvVar.CODE_API_KEY];
-        const { files, toolContext } = await primeCodeFiles(options, codeApiKey);
+        const { files, toolContext } = await primeCodeFiles(
+          {
+            ...options,
+            agentId: agent?.id,
+          },
+          codeApiKey,
+        );
        if (toolContext) {
          toolContextMap[tool] = toolContext;
        }
@@ -262,17 +261,48 @@ const loadTools = async ({
      continue;
    } else if (tool === Tools.file_search) {
      requestedTools[tool] = async () => {
-        const { files, toolContext } = await primeSearchFiles(options);
+        const { files, toolContext } = await primeSearchFiles({
+          ...options,
+          agentId: agent?.id,
+        });
        if (toolContext) {
          toolContextMap[tool] = toolContext;
        }
        return createFileSearchTool({ req: options.req, files, entity_id: agent?.id });
      };
      continue;
-    } else if (tool && appTools[tool] && mcpToolPattern.test(tool)) {
+    } else if (tool === Tools.web_search) {
+      const webSearchConfig = options?.req?.app?.locals?.webSearch;
+      const result = await loadWebSearchAuth({
+        userId: user,
+        loadAuthValues,
+        webSearchConfig,
+      });
+      const { onSearchResults, onGetHighlights } = options?.[Tools.web_search] ?? {};
+      requestedTools[tool] = async () => {
+        toolContextMap[tool] = `# \`${tool}\`:
+Current Date & Time: ${replaceSpecialVars({ text: '{{iso_datetime}}' })}
+1. **Execute immediately without preface** when using \`${tool}\`.
+2. **After the search, begin with a brief summary** that directly addresses the query without headers or explaining your process.
+3. **Structure your response clearly** using Markdown formatting (Level 2 headers for sections, lists for multiple points, tables for comparisons).
+4. **Cite sources properly** according to the citation anchor format, utilizing group anchors when appropriate.
+5. **Tailor your approach to the query type** (academic, news, coding, etc.) while maintaining an expert, journalistic, unbiased tone.
+6. **Provide comprehensive information** with specific details, examples, and as much relevant context as possible from search results.
+7. **Avoid moralizing language.**
+`.trim();
+        return createSearchTool({
+          ...result.authResult,
+          onSearchResults,
+          onGetHighlights,
+          logger,
+        });
+      };
+      continue;
+    } else if (tool && cachedTools && mcpToolPattern.test(tool)) {
      requestedTools[tool] = async () =>
        createMCPTool({
          req: options.req,
+          res: options.res,
          toolKey: tool,
          model: agent?.model ?? model,
          provider: agent?.provider ?? endpoint,
@@ -281,7 +311,7 @@ const loadTools = async ({
    }

    if (customConstructors[tool]) {
-      requestedTools[tool] = customConstructors[tool];
+      requestedTools[tool] = async () => customConstructors[tool](toolContextMap);
      continue;
    }

@@ -296,30 +326,6 @@ const loadTools = async ({
      requestedTools[tool] = toolInstance;
      continue;
    }
-
-    if (functions === true) {
-      remainingTools.push(tool);
-    }
-  }
-
-  let specs = null;
-  if (useSpecs === true && functions === true && remainingTools.length > 0) {
-    specs = await loadSpecs({
-      llm: model,
-      user,
-      message: options.message,
-      memory: options.memory,
-      signal: options.signal,
-      tools: remainingTools,
-      map: true,
-      verbose: false,
-    });
-  }
-
-  for (const tool of remainingTools) {
-    if (specs && specs[tool]) {
-      requestedTools[tool] = specs[tool];
-    }
  }

  if (returnMap) {
@@ -345,7 +351,6 @@ const loadTools = async ({

 module.exports = {
  loadToolWithAuth,
-  loadAuthValues,
  validateTools,
  loadTools,
 };
--- a/api/app/clients/tools/util/handleTools.test.js
+++ b/api/app/clients/tools/util/handleTools.test.js
@@ -1,8 +1,5 @@
-const mockUser = {
-  _id: 'fakeId',
-  save: jest.fn(),
-  findByIdAndDelete: jest.fn(),
-};
+const mongoose = require('mongoose');
+const { MongoMemoryServer } = require('mongodb-memory-server');

 const mockPluginService = {
  updateUserPluginAuth: jest.fn(),
@@ -10,23 +7,18 @@ const mockPluginService = {
  getUserPluginAuthValue: jest.fn(),
 };

-jest.mock('~/models/User', () => {
-  return function () {
-    return mockUser;
-  };
-});
-
 jest.mock('~/server/services/PluginService', () => mockPluginService);

 const { BaseLLM } = require('@langchain/openai');
 const { Calculator } = require('@langchain/community/tools/calculator');

-const User = require('~/models/User');
+const { User } = require('~/db/models');
 const PluginService = require('~/server/services/PluginService');
 const { validateTools, loadTools, loadToolWithAuth } = require('./handleTools');
 const { StructuredSD, availableTools, DALLE3 } = require('../');

 describe('Tool Handlers', () => {
+  let mongoServer;
  let fakeUser;
  const pluginKey = 'dalle';
  const pluginKey2 = 'wolfram';
@@ -37,7 +29,9 @@ describe('Tool Handlers', () => {
  const authConfigs = mainPlugin.authConfig;

  beforeAll(async () => {
-    mockUser.save.mockResolvedValue(undefined);
+    mongoServer = await MongoMemoryServer.create();
+    const mongoUri = mongoServer.getUri();
+    await mongoose.connect(mongoUri);

    const userAuthValues = {};
    mockPluginService.getUserPluginAuthValue.mockImplementation((userId, authField) => {
@@ -78,9 +72,36 @@ describe('Tool Handlers', () => {
  });

  afterAll(async () => {
-    await mockUser.findByIdAndDelete(fakeUser._id);
+    await mongoose.disconnect();
+    await mongoServer.stop();
+  });
+
+  beforeEach(async () => {
+    // Clear mocks but not the database since we need the user to persist
+    jest.clearAllMocks();
+
+    // Reset the mock implementations
+    const userAuthValues = {};
+    mockPluginService.getUserPluginAuthValue.mockImplementation((userId, authField) => {
+      return userAuthValues[`${userId}-${authField}`];
+    });
+    mockPluginService.updateUserPluginAuth.mockImplementation(
+      (userId, authField, _pluginKey, credential) => {
+        const fields = authField.split('||');
+        fields.forEach((field) => {
+          userAuthValues[`${userId}-${field}`] = credential;
+        });
+      },
+    );
+
+    // Re-add the auth configs for the user
    for (const authConfig of authConfigs) {
-      await PluginService.deleteUserPluginAuth(fakeUser._id, authConfig.authField);
+      await PluginService.updateUserPluginAuth(
+        fakeUser._id,
+        authConfig.authField,
+        pluginKey,
+        mockCredential,
+      );
    }
  });

@@ -218,7 +239,6 @@ describe('Tool Handlers', () => {
      try {
        await loadTool2();
      } catch (error) {
-        // eslint-disable-next-line jest/no-conditional-expect
        expect(error).toBeDefined();
      }
    });
--- a/api/app/clients/tools/util/index.js
+++ b/api/app/clients/tools/util/index.js
@@ -1,9 +1,8 @@
-const { validateTools, loadTools, loadAuthValues } = require('./handleTools');
+const { validateTools, loadTools } = require('./handleTools');
 const handleOpenAIErrors = require('./handleOpenAIErrors');

 module.exports = {
  handleOpenAIErrors,
-  loadAuthValues,
  validateTools,
  loadTools,
 };
--- a/api/app/clients/tools/util/loadSpecs.js
+++ b/api/app/clients/tools/util/loadSpecs.js
@@ -1,117 +0,0 @@
-const fs = require('fs');
-const path = require('path');
-const { z } = require('zod');
-const { logger } = require('~/config');
-const { createOpenAPIPlugin } = require('~/app/clients/tools/dynamic/OpenAPIPlugin');
-
-// The minimum Manifest definition
-const ManifestDefinition = z.object({
-  schema_version: z.string().optional(),
-  name_for_human: z.string(),
-  name_for_model: z.string(),
-  description_for_human: z.string(),
-  description_for_model: z.string(),
-  auth: z.object({}).optional(),
-  api: z.object({
-    // Spec URL or can be the filename of the OpenAPI spec yaml file,
-    // located in api\app\clients\tools\.well-known\openapi
-    url: z.string(),
-    type: z.string().optional(),
-    is_user_authenticated: z.boolean().nullable().optional(),
-    has_user_authentication: z.boolean().nullable().optional(),
-  }),
-  // use to override any params that the LLM will consistently get wrong
-  params: z.object({}).optional(),
-  logo_url: z.string().optional(),
-  contact_email: z.string().optional(),
-  legal_info_url: z.string().optional(),
-});
-
-function validateJson(json) {
-  try {
-    return ManifestDefinition.parse(json);
-  } catch (error) {
-    logger.debug('[validateJson] manifest parsing error', error);
-    return false;
-  }
-}
-
-// omit the LLM to return the well known jsons as objects
-async function loadSpecs({ llm, user, message, tools = [], map = false, memory, signal }) {
-  const directoryPath = path.join(__dirname, '..', '.well-known');
-  let files = [];
-
-  for (let i = 0; i < tools.length; i++) {
-    const filePath = path.join(directoryPath, tools[i] + '.json');
-
-    try {
-      // If the access Promise is resolved, it means that the file exists
-      // Then we can add it to the files array
-      await fs.promises.access(filePath, fs.constants.F_OK);
-      files.push(tools[i] + '.json');
-    } catch (err) {
-      logger.error(`[loadSpecs] File ${tools[i] + '.json'} does not exist`, err);
-    }
-  }
-
-  if (files.length === 0) {
-    files = (await fs.promises.readdir(directoryPath)).filter(
-      (file) => path.extname(file) === '.json',
-    );
-  }
-
-  const validJsons = [];
-  const constructorMap = {};
-
-  logger.debug('[validateJson] files', files);
-
-  for (const file of files) {
-    if (path.extname(file) === '.json') {
-      const filePath = path.join(directoryPath, file);
-      const fileContent = await fs.promises.readFile(filePath, 'utf8');
-      const json = JSON.parse(fileContent);
-
-      if (!validateJson(json)) {
-        logger.debug('[validateJson] Invalid json', json);
-        continue;
-      }
-
-      if (llm && map) {
-        constructorMap[json.name_for_model] = async () =>
-          await createOpenAPIPlugin({
-            data: json,
-            llm,
-            message,
-            memory,
-            signal,
-            user,
-          });
-        continue;
-      }
-
-      if (llm) {
-        validJsons.push(createOpenAPIPlugin({ data: json, llm }));
-        continue;
-      }
-
-      validJsons.push(json);
-    }
-  }
-
-  if (map) {
-    return constructorMap;
-  }
-
-  const plugins = (await Promise.all(validJsons)).filter((plugin) => plugin);
-
-  //   logger.debug('[validateJson] plugins', plugins);
-  //   logger.debug(plugins[0].name);
-
-  return plugins;
-}
-
-module.exports = {
-  loadSpecs,
-  validateJson,
-  ManifestDefinition,
-};
--- a/api/app/clients/tools/util/loadSpecs.spec.js
+++ b/api/app/clients/tools/util/loadSpecs.spec.js
@@ -1,101 +0,0 @@
-const fs = require('fs');
-const { validateJson, loadSpecs, ManifestDefinition } = require('./loadSpecs');
-const { createOpenAPIPlugin } = require('../dynamic/OpenAPIPlugin');
-
-jest.mock('../dynamic/OpenAPIPlugin');
-
-describe('ManifestDefinition', () => {
-  it('should validate correct json', () => {
-    const json = {
-      name_for_human: 'Test',
-      name_for_model: 'Test',
-      description_for_human: 'Test',
-      description_for_model: 'Test',
-      api: {
-        url: 'http://test.com',
-      },
-    };
-
-    expect(() => ManifestDefinition.parse(json)).not.toThrow();
-  });
-
-  it('should not validate incorrect json', () => {
-    const json = {
-      name_for_human: 'Test',
-      name_for_model: 'Test',
-      description_for_human: 'Test',
-      description_for_model: 'Test',
-      api: {
-        url: 123, // incorrect type
-      },
-    };
-
-    expect(() => ManifestDefinition.parse(json)).toThrow();
-  });
-});
-
-describe('validateJson', () => {
-  it('should return parsed json if valid', () => {
-    const json = {
-      name_for_human: 'Test',
-      name_for_model: 'Test',
-      description_for_human: 'Test',
-      description_for_model: 'Test',
-      api: {
-        url: 'http://test.com',
-      },
-    };
-
-    expect(validateJson(json)).toEqual(json);
-  });
-
-  it('should return false if json is not valid', () => {
-    const json = {
-      name_for_human: 'Test',
-      name_for_model: 'Test',
-      description_for_human: 'Test',
-      description_for_model: 'Test',
-      api: {
-        url: 123, // incorrect type
-      },
-    };
-
-    expect(validateJson(json)).toEqual(false);
-  });
-});
-
-describe('loadSpecs', () => {
-  beforeEach(() => {
-    jest.spyOn(fs.promises, 'readdir').mockResolvedValue(['test.json']);
-    jest.spyOn(fs.promises, 'readFile').mockResolvedValue(
-      JSON.stringify({
-        name_for_human: 'Test',
-        name_for_model: 'Test',
-        description_for_human: 'Test',
-        description_for_model: 'Test',
-        api: {
-          url: 'http://test.com',
-        },
-      }),
-    );
-    createOpenAPIPlugin.mockResolvedValue({});
-  });
-
-  afterEach(() => {
-    jest.restoreAllMocks();
-  });
-
-  it('should return plugins', async () => {
-    const plugins = await loadSpecs({ llm: true, verbose: false });
-
-    expect(plugins).toHaveLength(1);
-    expect(createOpenAPIPlugin).toHaveBeenCalledTimes(1);
-  });
-
-  it('should return constructorMap if map is true', async () => {
-    const plugins = await loadSpecs({ llm: {}, map: true, verbose: false });
-
-    expect(plugins).toHaveProperty('Test');
-    expect(createOpenAPIPlugin).not.toHaveBeenCalled();
-  });
-});
--- a/api/cache/banViolation.js
+++ b/api/cache/banViolation.js
@@ -1,8 +1,9 @@
+const { logger } = require('@librechat/data-schemas');
+const { isEnabled, math } = require('@librechat/api');
 const { ViolationTypes } = require('librechat-data-provider');
-const { isEnabled, math, removePorts } = require('~/server/utils');
 const { deleteAllUserSessions } = require('~/models');
+const { removePorts } = require('~/server/utils');
 const getLogStores = require('./getLogStores');
-const { logger } = require('~/config');

 const { BAN_VIOLATIONS, BAN_INTERVAL } = process.env ?? {};
 const interval = math(BAN_INTERVAL, 20);
@@ -32,7 +33,6 @@ const banViolation = async (req, res, errorMessage) => {
  if (!isEnabled(BAN_VIOLATIONS)) {
    return;
  }
-
  if (!errorMessage) {
    return;
  }
@@ -51,7 +51,6 @@ const banViolation = async (req, res, errorMessage) => {

  const banLogs = getLogStores(ViolationTypes.BAN);
  const duration = errorMessage.duration || banLogs.opts.ttl;
-
  if (duration <= 0) {
    return;
  }
--- a/api/cache/banViolation.spec.js
+++ b/api/cache/banViolation.spec.js
@@ -1,48 +1,28 @@
+const mongoose = require('mongoose');
+const { MongoMemoryServer } = require('mongodb-memory-server');
 const banViolation = require('./banViolation');

-jest.mock('keyv');
-jest.mock('../models/Session');
-// Mocking the getLogStores function
-jest.mock('./getLogStores', () => {
-  return jest.fn().mockImplementation(() => {
-    const EventEmitter = require('events');
-    const { CacheKeys } = require('librechat-data-provider');
-    const math = require('../server/utils/math');
-    const mockGet = jest.fn();
-    const mockSet = jest.fn();
-    class KeyvMongo extends EventEmitter {
-      constructor(url = 'mongodb://127.0.0.1:27017', options) {
-        super();
-        this.ttlSupport = false;
-        url = url ?? {};
-        if (typeof url === 'string') {
-          url = { url };
-        }
-        if (url.uri) {
-          url = { url: url.uri, ...url };
-        }
-        this.opts = {
-          url,
-          collection: 'keyv',
-          ...url,
-          ...options,
-        };
-      }
-
-      get = mockGet;
-      set = mockSet;
-    }
-
-    return new KeyvMongo('', {
-      namespace: CacheKeys.BANS,
-      ttl: math(process.env.BAN_DURATION, 7200000),
-    });
-  });
-});
+// Mock deleteAllUserSessions since we're testing ban logic, not session deletion
+jest.mock('~/models', () => ({
+  ...jest.requireActual('~/models'),
+  deleteAllUserSessions: jest.fn().mockResolvedValue(true),
+}));

 describe('banViolation', () => {
+  let mongoServer;
  let req, res, errorMessage;

+  beforeAll(async () => {
+    mongoServer = await MongoMemoryServer.create();
+    const mongoUri = mongoServer.getUri();
+    await mongoose.connect(mongoUri);
+  });
+
+  afterAll(async () => {
+    await mongoose.disconnect();
+    await mongoServer.stop();
+  });
+
  beforeEach(() => {
    req = {
      ip: '127.0.0.1',
@@ -55,7 +35,7 @@ describe('banViolation', () => {
    };
    errorMessage = {
      type: 'someViolation',
-      user_id: '12345',
+      user_id: new mongoose.Types.ObjectId().toString(), // Use valid ObjectId
      prev_count: 0,
      violation_count: 0,
    };
--- a/api/cache/cacheConfig.js
+++ b/api/cache/cacheConfig.js
@@ -0,0 +1,54 @@
+const fs = require('fs');
+const { math, isEnabled } = require('@librechat/api');
+const { CacheKeys } = require('librechat-data-provider');
+
+// To ensure that different deployments do not interfere with each other's cache, we use a prefix for the Redis keys.
+// This prefix is usually the deployment ID, which is often passed to the container or pod as an env var.
+// Set REDIS_KEY_PREFIX_VAR to the env var that contains the deployment ID.
+const REDIS_KEY_PREFIX_VAR = process.env.REDIS_KEY_PREFIX_VAR;
+const REDIS_KEY_PREFIX = process.env.REDIS_KEY_PREFIX;
+if (REDIS_KEY_PREFIX_VAR && REDIS_KEY_PREFIX) {
+  throw new Error('Only either REDIS_KEY_PREFIX_VAR or REDIS_KEY_PREFIX can be set.');
+}
+
+const USE_REDIS = isEnabled(process.env.USE_REDIS);
+if (USE_REDIS && !process.env.REDIS_URI) {
+  throw new Error('USE_REDIS is enabled but REDIS_URI is not set.');
+}
+
+// Comma-separated list of cache namespaces that should be forced to use in-memory storage
+// even when Redis is enabled. This allows selective performance optimization for specific caches.
+const FORCED_IN_MEMORY_CACHE_NAMESPACES = process.env.FORCED_IN_MEMORY_CACHE_NAMESPACES
+  ? process.env.FORCED_IN_MEMORY_CACHE_NAMESPACES.split(',').map((key) => key.trim())
+  : [];
+
+// Validate against CacheKeys enum
+if (FORCED_IN_MEMORY_CACHE_NAMESPACES.length > 0) {
+  const validKeys = Object.values(CacheKeys);
+  const invalidKeys = FORCED_IN_MEMORY_CACHE_NAMESPACES.filter((key) => !validKeys.includes(key));
+
+  if (invalidKeys.length > 0) {
+    throw new Error(
+      `Invalid cache keys in FORCED_IN_MEMORY_CACHE_NAMESPACES: ${invalidKeys.join(', ')}. Valid keys: ${validKeys.join(', ')}`,
+    );
+  }
+}
+
+const cacheConfig = {
+  FORCED_IN_MEMORY_CACHE_NAMESPACES,
+  USE_REDIS,
+  REDIS_URI: process.env.REDIS_URI,
+  REDIS_USERNAME: process.env.REDIS_USERNAME,
+  REDIS_PASSWORD: process.env.REDIS_PASSWORD,
+  REDIS_CA: process.env.REDIS_CA ? fs.readFileSync(process.env.REDIS_CA, 'utf8') : null,
+  REDIS_KEY_PREFIX: process.env[REDIS_KEY_PREFIX_VAR] || REDIS_KEY_PREFIX || '',
+  REDIS_MAX_LISTENERS: math(process.env.REDIS_MAX_LISTENERS, 40),
+  REDIS_PING_INTERVAL: math(process.env.REDIS_PING_INTERVAL, 0),
+
+  CI: isEnabled(process.env.CI),
+  DEBUG_MEMORY_CACHE: isEnabled(process.env.DEBUG_MEMORY_CACHE),
+
+  BAN_DURATION: math(process.env.BAN_DURATION, 7200000), // 2 hours
+};
+
+module.exports = { cacheConfig };
--- a/api/cache/cacheConfig.spec.js
+++ b/api/cache/cacheConfig.spec.js
@@ -0,0 +1,157 @@
+const fs = require('fs');
+
+describe('cacheConfig', () => {
+  let originalEnv;
+  let originalReadFileSync;
+
+  beforeEach(() => {
+    originalEnv = { ...process.env };
+    originalReadFileSync = fs.readFileSync;
+
+    // Clear all related env vars first
+    delete process.env.REDIS_URI;
+    delete process.env.REDIS_CA;
+    delete process.env.REDIS_KEY_PREFIX_VAR;
+    delete process.env.REDIS_KEY_PREFIX;
+    delete process.env.USE_REDIS;
+    delete process.env.REDIS_PING_INTERVAL;
+    delete process.env.FORCED_IN_MEMORY_CACHE_NAMESPACES;
+
+    // Clear require cache
+    jest.resetModules();
+  });
+
+  afterEach(() => {
+    process.env = originalEnv;
+    fs.readFileSync = originalReadFileSync;
+    jest.resetModules();
+  });
+
+  describe('REDIS_KEY_PREFIX validation and resolution', () => {
+    test('should throw error when both REDIS_KEY_PREFIX_VAR and REDIS_KEY_PREFIX are set', () => {
+      process.env.REDIS_KEY_PREFIX_VAR = 'DEPLOYMENT_ID';
+      process.env.REDIS_KEY_PREFIX = 'manual-prefix';
+
+      expect(() => {
+        require('./cacheConfig');
+      }).toThrow('Only either REDIS_KEY_PREFIX_VAR or REDIS_KEY_PREFIX can be set.');
+    });
+
+    test('should resolve REDIS_KEY_PREFIX from variable reference', () => {
+      process.env.REDIS_KEY_PREFIX_VAR = 'DEPLOYMENT_ID';
+      process.env.DEPLOYMENT_ID = 'test-deployment-123';
+
+      const { cacheConfig } = require('./cacheConfig');
+      expect(cacheConfig.REDIS_KEY_PREFIX).toBe('test-deployment-123');
+    });
+
+    test('should use direct REDIS_KEY_PREFIX value', () => {
+      process.env.REDIS_KEY_PREFIX = 'direct-prefix';
+
+      const { cacheConfig } = require('./cacheConfig');
+      expect(cacheConfig.REDIS_KEY_PREFIX).toBe('direct-prefix');
+    });
+
+    test('should default to empty string when no prefix is configured', () => {
+      const { cacheConfig } = require('./cacheConfig');
+      expect(cacheConfig.REDIS_KEY_PREFIX).toBe('');
+    });
+
+    test('should handle empty variable reference', () => {
+      process.env.REDIS_KEY_PREFIX_VAR = 'EMPTY_VAR';
+      process.env.EMPTY_VAR = '';
+
+      const { cacheConfig } = require('./cacheConfig');
+      expect(cacheConfig.REDIS_KEY_PREFIX).toBe('');
+    });
+
+    test('should handle undefined variable reference', () => {
+      process.env.REDIS_KEY_PREFIX_VAR = 'UNDEFINED_VAR';
+
+      const { cacheConfig } = require('./cacheConfig');
+      expect(cacheConfig.REDIS_KEY_PREFIX).toBe('');
+    });
+  });
+
+  describe('USE_REDIS and REDIS_URI validation', () => {
+    test('should throw error when USE_REDIS is enabled but REDIS_URI is not set', () => {
+      process.env.USE_REDIS = 'true';
+
+      expect(() => {
+        require('./cacheConfig');
+      }).toThrow('USE_REDIS is enabled but REDIS_URI is not set.');
+    });
+
+    test('should not throw error when USE_REDIS is enabled and REDIS_URI is set', () => {
+      process.env.USE_REDIS = 'true';
+      process.env.REDIS_URI = 'redis://localhost:6379';
+
+      expect(() => {
+        require('./cacheConfig');
+      }).not.toThrow();
+    });
+
+    test('should handle empty REDIS_URI when USE_REDIS is enabled', () => {
+      process.env.USE_REDIS = 'true';
+      process.env.REDIS_URI = '';
+
+      expect(() => {
+        require('./cacheConfig');
+      }).toThrow('USE_REDIS is enabled but REDIS_URI is not set.');
+    });
+  });
+
+  describe('REDIS_CA file reading', () => {
+    test('should be null when REDIS_CA is not set', () => {
+      const { cacheConfig } = require('./cacheConfig');
+      expect(cacheConfig.REDIS_CA).toBeNull();
+    });
+  });
+
+  describe('REDIS_PING_INTERVAL configuration', () => {
+    test('should default to 0 when REDIS_PING_INTERVAL is not set', () => {
+      const { cacheConfig } = require('./cacheConfig');
+      expect(cacheConfig.REDIS_PING_INTERVAL).toBe(0);
+    });
+
+    test('should use provided REDIS_PING_INTERVAL value', () => {
+      process.env.REDIS_PING_INTERVAL = '300';
+
+      const { cacheConfig } = require('./cacheConfig');
+      expect(cacheConfig.REDIS_PING_INTERVAL).toBe(300);
+    });
+  });
+
+  describe('FORCED_IN_MEMORY_CACHE_NAMESPACES validation', () => {
+    test('should parse comma-separated cache keys correctly', () => {
+      process.env.FORCED_IN_MEMORY_CACHE_NAMESPACES = ' ROLES, STATIC_CONFIG ,MESSAGES ';
+
+      const { cacheConfig } = require('./cacheConfig');
+      expect(cacheConfig.FORCED_IN_MEMORY_CACHE_NAMESPACES).toEqual([
+        'ROLES',
+        'STATIC_CONFIG',
+        'MESSAGES',
+      ]);
+    });
+
+    test('should throw error for invalid cache keys', () => {
+      process.env.FORCED_IN_MEMORY_CACHE_NAMESPACES = 'INVALID_KEY,ROLES';
+
+      expect(() => {
+        require('./cacheConfig');
+      }).toThrow('Invalid cache keys in FORCED_IN_MEMORY_CACHE_NAMESPACES: INVALID_KEY');
+    });
+
+    test('should handle empty string gracefully', () => {
+      process.env.FORCED_IN_MEMORY_CACHE_NAMESPACES = '';
+
+      const { cacheConfig } = require('./cacheConfig');
+      expect(cacheConfig.FORCED_IN_MEMORY_CACHE_NAMESPACES).toEqual([]);
+    });
+
+    test('should handle undefined env var gracefully', () => {
+      const { cacheConfig } = require('./cacheConfig');
+      expect(cacheConfig.FORCED_IN_MEMORY_CACHE_NAMESPACES).toEqual([]);
+    });
+  });
+});
--- a/api/cache/cacheFactory.js
+++ b/api/cache/cacheFactory.js
@@ -0,0 +1,69 @@
+const KeyvRedis = require('@keyv/redis').default;
+const { Keyv } = require('keyv');
+const { cacheConfig } = require('./cacheConfig');
+const { keyvRedisClient, ioredisClient, GLOBAL_PREFIX_SEPARATOR } = require('./redisClients');
+const { Time } = require('librechat-data-provider');
+const { RedisStore: ConnectRedis } = require('connect-redis');
+const MemoryStore = require('memorystore')(require('express-session'));
+const { violationFile } = require('./keyvFiles');
+const { RedisStore } = require('rate-limit-redis');
+
+/**
+ * Creates a cache instance using Redis or a fallback store. Suitable for general caching needs.
+ * @param {string} namespace - The cache namespace.
+ * @param {number} [ttl] - Time to live for cache entries.
+ * @param {object} [fallbackStore] - Optional fallback store if Redis is not used.
+ * @returns {Keyv} Cache instance.
+ */
+const standardCache = (namespace, ttl = undefined, fallbackStore = undefined) => {
+  if (
+    cacheConfig.USE_REDIS &&
+    !cacheConfig.FORCED_IN_MEMORY_CACHE_NAMESPACES?.includes(namespace)
+  ) {
+    const keyvRedis = new KeyvRedis(keyvRedisClient);
+    const cache = new Keyv(keyvRedis, { namespace, ttl });
+    keyvRedis.namespace = cacheConfig.REDIS_KEY_PREFIX;
+    keyvRedis.keyPrefixSeparator = GLOBAL_PREFIX_SEPARATOR;
+    return cache;
+  }
+  if (fallbackStore) return new Keyv({ store: fallbackStore, namespace, ttl });
+  return new Keyv({ namespace, ttl });
+};
+
+/**
+ * Creates a cache instance for storing violation data.
+ * Uses a file-based fallback store if Redis is not enabled.
+ * @param {string} namespace - The cache namespace for violations.
+ * @param {number} [ttl] - Time to live for cache entries.
+ * @returns {Keyv} Cache instance for violations.
+ */
+const violationCache = (namespace, ttl = undefined) => {
+  return standardCache(`violations:${namespace}`, ttl, violationFile);
+};
+
+/**
+ * Creates a session cache instance using Redis or in-memory store.
+ * @param {string} namespace - The session namespace.
+ * @param {number} [ttl] - Time to live for session entries.
+ * @returns {MemoryStore | ConnectRedis} Session store instance.
+ */
+const sessionCache = (namespace, ttl = undefined) => {
+  namespace = namespace.endsWith(':') ? namespace : `${namespace}:`;
+  if (!cacheConfig.USE_REDIS) return new MemoryStore({ ttl, checkPeriod: Time.ONE_DAY });
+  return new ConnectRedis({ client: ioredisClient, ttl, prefix: namespace });
+};
+
+/**
+ * Creates a rate limiter cache using Redis.
+ * @param {string} prefix - The key prefix for rate limiting.
+ * @returns {RedisStore|undefined} RedisStore instance or undefined if Redis is not used.
+ */
+const limiterCache = (prefix) => {
+  if (!prefix) throw new Error('prefix is required');
+  if (!cacheConfig.USE_REDIS) return undefined;
+  prefix = prefix.endsWith(':') ? prefix : `${prefix}:`;
+  return new RedisStore({ sendCommand, prefix });
+};
+const sendCommand = (...args) => ioredisClient?.call(...args);
+
+module.exports = { standardCache, sessionCache, violationCache, limiterCache };
--- a/api/cache/cacheFactory.spec.js
+++ b/api/cache/cacheFactory.spec.js
@@ -0,0 +1,296 @@
+const { Time } = require('librechat-data-provider');
+
+// Mock dependencies first
+const mockKeyvRedis = {
+  namespace: '',
+  keyPrefixSeparator: '',
+};
+
+const mockKeyv = jest.fn().mockReturnValue({ mock: 'keyv' });
+const mockConnectRedis = jest.fn().mockReturnValue({ mock: 'connectRedis' });
+const mockMemoryStore = jest.fn().mockReturnValue({ mock: 'memoryStore' });
+const mockRedisStore = jest.fn().mockReturnValue({ mock: 'redisStore' });
+
+const mockIoredisClient = {
+  call: jest.fn(),
+};
+
+const mockKeyvRedisClient = {};
+const mockViolationFile = {};
+
+// Mock modules before requiring the main module
+jest.mock('@keyv/redis', () => ({
+  default: jest.fn().mockImplementation(() => mockKeyvRedis),
+}));
+
+jest.mock('keyv', () => ({
+  Keyv: mockKeyv,
+}));
+
+jest.mock('./cacheConfig', () => ({
+  cacheConfig: {
+    USE_REDIS: false,
+    REDIS_KEY_PREFIX: 'test',
+    FORCED_IN_MEMORY_CACHE_NAMESPACES: [],
+  },
+}));
+
+jest.mock('./redisClients', () => ({
+  keyvRedisClient: mockKeyvRedisClient,
+  ioredisClient: mockIoredisClient,
+  GLOBAL_PREFIX_SEPARATOR: '::',
+}));
+
+jest.mock('./keyvFiles', () => ({
+  violationFile: mockViolationFile,
+}));
+
+jest.mock('connect-redis', () => ({ RedisStore: mockConnectRedis }));
+
+jest.mock('memorystore', () => jest.fn(() => mockMemoryStore));
+
+jest.mock('rate-limit-redis', () => ({
+  RedisStore: mockRedisStore,
+}));
+
+// Import after mocking
+const { standardCache, sessionCache, violationCache, limiterCache } = require('./cacheFactory');
+const { cacheConfig } = require('./cacheConfig');
+
+describe('cacheFactory', () => {
+  beforeEach(() => {
+    jest.clearAllMocks();
+
+    // Reset cache config mock
+    cacheConfig.USE_REDIS = false;
+    cacheConfig.REDIS_KEY_PREFIX = 'test';
+    cacheConfig.FORCED_IN_MEMORY_CACHE_NAMESPACES = [];
+  });
+
+  describe('redisCache', () => {
+    it('should create Redis cache when USE_REDIS is true', () => {
+      cacheConfig.USE_REDIS = true;
+      const namespace = 'test-namespace';
+      const ttl = 3600;
+
+      standardCache(namespace, ttl);
+
+      expect(require('@keyv/redis').default).toHaveBeenCalledWith(mockKeyvRedisClient);
+      expect(mockKeyv).toHaveBeenCalledWith(mockKeyvRedis, { namespace, ttl });
+      expect(mockKeyvRedis.namespace).toBe(cacheConfig.REDIS_KEY_PREFIX);
+      expect(mockKeyvRedis.keyPrefixSeparator).toBe('::');
+    });
+
+    it('should create Redis cache with undefined ttl when not provided', () => {
+      cacheConfig.USE_REDIS = true;
+      const namespace = 'test-namespace';
+
+      standardCache(namespace);
+
+      expect(mockKeyv).toHaveBeenCalledWith(mockKeyvRedis, { namespace, ttl: undefined });
+    });
+
+    it('should use fallback store when USE_REDIS is false and fallbackStore is provided', () => {
+      cacheConfig.USE_REDIS = false;
+      const namespace = 'test-namespace';
+      const ttl = 3600;
+      const fallbackStore = { some: 'store' };
+
+      standardCache(namespace, ttl, fallbackStore);
+
+      expect(mockKeyv).toHaveBeenCalledWith({ store: fallbackStore, namespace, ttl });
+    });
+
+    it('should create default Keyv instance when USE_REDIS is false and no fallbackStore', () => {
+      cacheConfig.USE_REDIS = false;
+      const namespace = 'test-namespace';
+      const ttl = 3600;
+
+      standardCache(namespace, ttl);
+
+      expect(mockKeyv).toHaveBeenCalledWith({ namespace, ttl });
+    });
+
+    it('should handle namespace and ttl as undefined', () => {
+      cacheConfig.USE_REDIS = false;
+
+      standardCache();
+
+      expect(mockKeyv).toHaveBeenCalledWith({ namespace: undefined, ttl: undefined });
+    });
+
+    it('should use fallback when namespace is in FORCED_IN_MEMORY_CACHE_NAMESPACES', () => {
+      cacheConfig.USE_REDIS = true;
+      cacheConfig.FORCED_IN_MEMORY_CACHE_NAMESPACES = ['forced-memory'];
+      const namespace = 'forced-memory';
+      const ttl = 3600;
+
+      standardCache(namespace, ttl);
+
+      expect(require('@keyv/redis').default).not.toHaveBeenCalled();
+      expect(mockKeyv).toHaveBeenCalledWith({ namespace, ttl });
+    });
+
+    it('should use Redis when namespace is not in FORCED_IN_MEMORY_CACHE_NAMESPACES', () => {
+      cacheConfig.USE_REDIS = true;
+      cacheConfig.FORCED_IN_MEMORY_CACHE_NAMESPACES = ['other-namespace'];
+      const namespace = 'test-namespace';
+      const ttl = 3600;
+
+      standardCache(namespace, ttl);
+
+      expect(require('@keyv/redis').default).toHaveBeenCalledWith(mockKeyvRedisClient);
+      expect(mockKeyv).toHaveBeenCalledWith(mockKeyvRedis, { namespace, ttl });
+    });
+  });
+
+  describe('violationCache', () => {
+    it('should create violation cache with prefixed namespace', () => {
+      const namespace = 'test-violations';
+      const ttl = 7200;
+
+      // We can't easily mock the internal redisCache call since it's in the same module
+      // But we can test that the function executes without throwing
+      expect(() => violationCache(namespace, ttl)).not.toThrow();
+    });
+
+    it('should create violation cache with undefined ttl', () => {
+      const namespace = 'test-violations';
+
+      violationCache(namespace);
+
+      // The function should call redisCache with violations: prefixed namespace
+      // Since we can't easily mock the internal redisCache call, we test the behavior
+      expect(() => violationCache(namespace)).not.toThrow();
+    });
+
+    it('should handle undefined namespace', () => {
+      expect(() => violationCache(undefined)).not.toThrow();
+    });
+  });
+
+  describe('sessionCache', () => {
+    it('should return MemoryStore when USE_REDIS is false', () => {
+      cacheConfig.USE_REDIS = false;
+      const namespace = 'sessions';
+      const ttl = 86400;
+
+      const result = sessionCache(namespace, ttl);
+
+      expect(mockMemoryStore).toHaveBeenCalledWith({ ttl, checkPeriod: Time.ONE_DAY });
+      expect(result).toBe(mockMemoryStore());
+    });
+
+    it('should return ConnectRedis when USE_REDIS is true', () => {
+      cacheConfig.USE_REDIS = true;
+      const namespace = 'sessions';
+      const ttl = 86400;
+
+      const result = sessionCache(namespace, ttl);
+
+      expect(mockConnectRedis).toHaveBeenCalledWith({
+        client: mockIoredisClient,
+        ttl,
+        prefix: `${namespace}:`,
+      });
+      expect(result).toBe(mockConnectRedis());
+    });
+
+    it('should add colon to namespace if not present', () => {
+      cacheConfig.USE_REDIS = true;
+      const namespace = 'sessions';
+
+      sessionCache(namespace);
+
+      expect(mockConnectRedis).toHaveBeenCalledWith({
+        client: mockIoredisClient,
+        ttl: undefined,
+        prefix: 'sessions:',
+      });
+    });
+
+    it('should not add colon to namespace if already present', () => {
+      cacheConfig.USE_REDIS = true;
+      const namespace = 'sessions:';
+
+      sessionCache(namespace);
+
+      expect(mockConnectRedis).toHaveBeenCalledWith({
+        client: mockIoredisClient,
+        ttl: undefined,
+        prefix: 'sessions:',
+      });
+    });
+
+    it('should handle undefined ttl', () => {
+      cacheConfig.USE_REDIS = false;
+      const namespace = 'sessions';
+
+      sessionCache(namespace);
+
+      expect(mockMemoryStore).toHaveBeenCalledWith({
+        ttl: undefined,
+        checkPeriod: Time.ONE_DAY,
+      });
+    });
+  });
+
+  describe('limiterCache', () => {
+    it('should return undefined when USE_REDIS is false', () => {
+      cacheConfig.USE_REDIS = false;
+      const result = limiterCache('prefix');
+
+      expect(result).toBeUndefined();
+    });
+
+    it('should return RedisStore when USE_REDIS is true', () => {
+      cacheConfig.USE_REDIS = true;
+      const result = limiterCache('rate-limit');
+
+      expect(mockRedisStore).toHaveBeenCalledWith({
+        sendCommand: expect.any(Function),
+        prefix: `rate-limit:`,
+      });
+      expect(result).toBe(mockRedisStore());
+    });
+
+    it('should add colon to prefix if not present', () => {
+      cacheConfig.USE_REDIS = true;
+      limiterCache('rate-limit');
+
+      expect(mockRedisStore).toHaveBeenCalledWith({
+        sendCommand: expect.any(Function),
+        prefix: 'rate-limit:',
+      });
+    });
+
+    it('should not add colon to prefix if already present', () => {
+      cacheConfig.USE_REDIS = true;
+      limiterCache('rate-limit:');
+
+      expect(mockRedisStore).toHaveBeenCalledWith({
+        sendCommand: expect.any(Function),
+        prefix: 'rate-limit:',
+      });
+    });
+
+    it('should pass sendCommand function that calls ioredisClient.call', () => {
+      cacheConfig.USE_REDIS = true;
+      limiterCache('rate-limit');
+
+      const sendCommandCall = mockRedisStore.mock.calls[0][0];
+      const sendCommand = sendCommandCall.sendCommand;
+
+      // Test that sendCommand properly delegates to ioredisClient.call
+      const args = ['GET', 'test-key'];
+      sendCommand(...args);
+
+      expect(mockIoredisClient.call).toHaveBeenCalledWith(...args);
+    });
+
+    it('should handle undefined prefix', () => {
+      cacheConfig.USE_REDIS = true;
+      expect(() => limiterCache()).toThrow('prefix is required');
+    });
+  });
+});
--- a/api/cache/clearPendingReq.js
+++ b/api/cache/clearPendingReq.js
@@ -1,7 +1,8 @@
+const { Time, CacheKeys } = require('librechat-data-provider');
+const { isEnabled } = require('~/server/utils');
 const getLogStores = require('./getLogStores');
-const { isEnabled } = require('../server/utils');
+
 const { USE_REDIS, LIMIT_CONCURRENT_MESSAGES } = process.env ?? {};
-const ttl = 1000 * 60 * 1;

 /**
 * Clear or decrement pending requests from the cache.
@@ -28,7 +29,7 @@ const clearPendingReq = async ({ userId, cache: _cache }) => {
    return;
  }

-  const namespace = 'pending_req';
+  const namespace = CacheKeys.PENDING_REQ;
  const cache = _cache ?? getLogStores(namespace);

  if (!cache) {
@@ -39,7 +40,7 @@ const clearPendingReq = async ({ userId, cache: _cache }) => {
  const currentReq = +((await cache.get(key)) ?? 0);

  if (currentReq && currentReq >= 1) {
-    await cache.set(key, currentReq - 1, ttl);
+    await cache.set(key, currentReq - 1, Time.ONE_MINUTE);
  } else {
    await cache.delete(key);
  }
--- a/api/cache/getLogStores.js
+++ b/api/cache/getLogStores.js
@@ -1,98 +1,53 @@
-const Keyv = require('keyv');
+const { cacheConfig } = require('./cacheConfig');
+const { Keyv } = require('keyv');
 const { CacheKeys, ViolationTypes, Time } = require('librechat-data-provider');
-const { logFile, violationFile } = require('./keyvFiles');
-const { math, isEnabled } = require('~/server/utils');
-const keyvRedis = require('./keyvRedis');
+const { logFile } = require('./keyvFiles');
 const keyvMongo = require('./keyvMongo');
-
-const { BAN_DURATION, USE_REDIS, DEBUG_MEMORY_CACHE, CI } = process.env ?? {};
-
-const duration = math(BAN_DURATION, 7200000);
-const isRedisEnabled = isEnabled(USE_REDIS);
-const debugMemoryCache = isEnabled(DEBUG_MEMORY_CACHE);
-
-const createViolationInstance = (namespace) => {
-  const config = isRedisEnabled ? { store: keyvRedis } : { store: violationFile, namespace };
-  return new Keyv(config);
-};
-
-// Serve cache from memory so no need to clear it on startup/exit
-const pending_req = isRedisEnabled
-  ? new Keyv({ store: keyvRedis })
-  : new Keyv({ namespace: 'pending_req' });
-
-const config = isRedisEnabled
-  ? new Keyv({ store: keyvRedis })
-  : new Keyv({ namespace: CacheKeys.CONFIG_STORE });
-
-const roles = isRedisEnabled
-  ? new Keyv({ store: keyvRedis })
-  : new Keyv({ namespace: CacheKeys.ROLES });
-
-const audioRuns = isRedisEnabled
-  ? new Keyv({ store: keyvRedis, ttl: Time.TEN_MINUTES })
-  : new Keyv({ namespace: CacheKeys.AUDIO_RUNS, ttl: Time.TEN_MINUTES });
-
-const messages = isRedisEnabled
-  ? new Keyv({ store: keyvRedis, ttl: Time.ONE_MINUTE })
-  : new Keyv({ namespace: CacheKeys.MESSAGES, ttl: Time.ONE_MINUTE });
-
-const flows = isRedisEnabled
-  ? new Keyv({ store: keyvRedis, ttl: Time.TWO_MINUTES })
-  : new Keyv({ namespace: CacheKeys.FLOWS, ttl: Time.ONE_MINUTE * 3 });
-
-const tokenConfig = isRedisEnabled
-  ? new Keyv({ store: keyvRedis, ttl: Time.THIRTY_MINUTES })
-  : new Keyv({ namespace: CacheKeys.TOKEN_CONFIG, ttl: Time.THIRTY_MINUTES });
-
-const genTitle = isRedisEnabled
-  ? new Keyv({ store: keyvRedis, ttl: Time.TWO_MINUTES })
-  : new Keyv({ namespace: CacheKeys.GEN_TITLE, ttl: Time.TWO_MINUTES });
-
-const modelQueries = isEnabled(process.env.USE_REDIS)
-  ? new Keyv({ store: keyvRedis })
-  : new Keyv({ namespace: CacheKeys.MODEL_QUERIES });
-
-const abortKeys = isRedisEnabled
-  ? new Keyv({ store: keyvRedis })
-  : new Keyv({ namespace: CacheKeys.ABORT_KEYS, ttl: Time.TEN_MINUTES });
+const { standardCache, sessionCache, violationCache } = require('./cacheFactory');

 const namespaces = {
-  [CacheKeys.ROLES]: roles,
-  [CacheKeys.CONFIG_STORE]: config,
-  pending_req,
-  [ViolationTypes.BAN]: new Keyv({ store: keyvMongo, namespace: CacheKeys.BANS, ttl: duration }),
-  [CacheKeys.ENCODED_DOMAINS]: new Keyv({
+  [ViolationTypes.GENERAL]: new Keyv({ store: logFile, namespace: 'violations' }),
+  [ViolationTypes.LOGINS]: violationCache(ViolationTypes.LOGINS),
+  [ViolationTypes.CONCURRENT]: violationCache(ViolationTypes.CONCURRENT),
+  [ViolationTypes.NON_BROWSER]: violationCache(ViolationTypes.NON_BROWSER),
+  [ViolationTypes.MESSAGE_LIMIT]: violationCache(ViolationTypes.MESSAGE_LIMIT),
+  [ViolationTypes.REGISTRATIONS]: violationCache(ViolationTypes.REGISTRATIONS),
+  [ViolationTypes.TOKEN_BALANCE]: violationCache(ViolationTypes.TOKEN_BALANCE),
+  [ViolationTypes.TTS_LIMIT]: violationCache(ViolationTypes.TTS_LIMIT),
+  [ViolationTypes.STT_LIMIT]: violationCache(ViolationTypes.STT_LIMIT),
+  [ViolationTypes.CONVO_ACCESS]: violationCache(ViolationTypes.CONVO_ACCESS),
+  [ViolationTypes.TOOL_CALL_LIMIT]: violationCache(ViolationTypes.TOOL_CALL_LIMIT),
+  [ViolationTypes.FILE_UPLOAD_LIMIT]: violationCache(ViolationTypes.FILE_UPLOAD_LIMIT),
+  [ViolationTypes.VERIFY_EMAIL_LIMIT]: violationCache(ViolationTypes.VERIFY_EMAIL_LIMIT),
+  [ViolationTypes.RESET_PASSWORD_LIMIT]: violationCache(ViolationTypes.RESET_PASSWORD_LIMIT),
+  [ViolationTypes.ILLEGAL_MODEL_REQUEST]: violationCache(ViolationTypes.ILLEGAL_MODEL_REQUEST),
+  [ViolationTypes.BAN]: new Keyv({
    store: keyvMongo,
-    namespace: CacheKeys.ENCODED_DOMAINS,
-    ttl: 0,
+    namespace: CacheKeys.BANS,
+    ttl: cacheConfig.BAN_DURATION,
  }),
-  general: new Keyv({ store: logFile, namespace: 'violations' }),
-  concurrent: createViolationInstance('concurrent'),
-  non_browser: createViolationInstance('non_browser'),
-  message_limit: createViolationInstance('message_limit'),
-  token_balance: createViolationInstance(ViolationTypes.TOKEN_BALANCE),
-  registrations: createViolationInstance('registrations'),
-  [ViolationTypes.TTS_LIMIT]: createViolationInstance(ViolationTypes.TTS_LIMIT),
-  [ViolationTypes.STT_LIMIT]: createViolationInstance(ViolationTypes.STT_LIMIT),
-  [ViolationTypes.CONVO_ACCESS]: createViolationInstance(ViolationTypes.CONVO_ACCESS),
-  [ViolationTypes.TOOL_CALL_LIMIT]: createViolationInstance(ViolationTypes.TOOL_CALL_LIMIT),
-  [ViolationTypes.FILE_UPLOAD_LIMIT]: createViolationInstance(ViolationTypes.FILE_UPLOAD_LIMIT),
-  [ViolationTypes.VERIFY_EMAIL_LIMIT]: createViolationInstance(ViolationTypes.VERIFY_EMAIL_LIMIT),
-  [ViolationTypes.RESET_PASSWORD_LIMIT]: createViolationInstance(
-    ViolationTypes.RESET_PASSWORD_LIMIT,
+
+  [CacheKeys.OPENID_SESSION]: sessionCache(CacheKeys.OPENID_SESSION),
+  [CacheKeys.SAML_SESSION]: sessionCache(CacheKeys.SAML_SESSION),
+
+  [CacheKeys.ROLES]: standardCache(CacheKeys.ROLES),
+  [CacheKeys.MCP_TOOLS]: standardCache(CacheKeys.MCP_TOOLS),
+  [CacheKeys.CONFIG_STORE]: standardCache(CacheKeys.CONFIG_STORE),
+  [CacheKeys.STATIC_CONFIG]: standardCache(CacheKeys.STATIC_CONFIG),
+  [CacheKeys.PENDING_REQ]: standardCache(CacheKeys.PENDING_REQ),
+  [CacheKeys.ENCODED_DOMAINS]: new Keyv({ store: keyvMongo, namespace: CacheKeys.ENCODED_DOMAINS }),
+  [CacheKeys.ABORT_KEYS]: standardCache(CacheKeys.ABORT_KEYS, Time.TEN_MINUTES),
+  [CacheKeys.TOKEN_CONFIG]: standardCache(CacheKeys.TOKEN_CONFIG, Time.THIRTY_MINUTES),
+  [CacheKeys.GEN_TITLE]: standardCache(CacheKeys.GEN_TITLE, Time.TWO_MINUTES),
+  [CacheKeys.S3_EXPIRY_INTERVAL]: standardCache(CacheKeys.S3_EXPIRY_INTERVAL, Time.THIRTY_MINUTES),
+  [CacheKeys.MODEL_QUERIES]: standardCache(CacheKeys.MODEL_QUERIES),
+  [CacheKeys.AUDIO_RUNS]: standardCache(CacheKeys.AUDIO_RUNS, Time.TEN_MINUTES),
+  [CacheKeys.MESSAGES]: standardCache(CacheKeys.MESSAGES, Time.ONE_MINUTE),
+  [CacheKeys.FLOWS]: standardCache(CacheKeys.FLOWS, Time.ONE_MINUTE * 3),
+  [CacheKeys.OPENID_EXCHANGED_TOKENS]: standardCache(
+    CacheKeys.OPENID_EXCHANGED_TOKENS,
+    Time.TEN_MINUTES,
  ),
-  [ViolationTypes.ILLEGAL_MODEL_REQUEST]: createViolationInstance(
-    ViolationTypes.ILLEGAL_MODEL_REQUEST,
-  ),
-  logins: createViolationInstance('logins'),
-  [CacheKeys.ABORT_KEYS]: abortKeys,
-  [CacheKeys.TOKEN_CONFIG]: tokenConfig,
-  [CacheKeys.GEN_TITLE]: genTitle,
-  [CacheKeys.MODEL_QUERIES]: modelQueries,
-  [CacheKeys.AUDIO_RUNS]: audioRuns,
-  [CacheKeys.MESSAGES]: messages,
-  [CacheKeys.FLOWS]: flows,
 };

 /**
@@ -101,7 +56,10 @@ const namespaces = {
 */
 function getTTLStores() {
  return Object.values(namespaces).filter(
-    (store) => store instanceof Keyv && typeof store.opts?.ttl === 'number' && store.opts.ttl > 0,
+    (store) =>
+      store instanceof Keyv &&
+      parseInt(store.opts?.ttl ?? '0') > 0 &&
+      !store.opts?.store?.constructor?.name?.includes('Redis'), // Only include non-Redis stores
  );
 }

@@ -137,18 +95,18 @@ async function clearExpiredFromCache(cache) {
      if (data?.expires && data.expires <= expiryTime) {
        const deleted = await cache.opts.store.delete(key);
        if (!deleted) {
-          debugMemoryCache &&
+          cacheConfig.DEBUG_MEMORY_CACHE &&
            console.warn(`[Cache] Error deleting entry: ${key} from ${cache.opts.namespace}`);
          continue;
        }
        cleared++;
      }
    } catch (error) {
-      debugMemoryCache &&
+      cacheConfig.DEBUG_MEMORY_CACHE &&
        console.log(`[Cache] Error processing entry from ${cache.opts.namespace}:`, error);
      const deleted = await cache.opts.store.delete(key);
      if (!deleted) {
-        debugMemoryCache &&
+        cacheConfig.DEBUG_MEMORY_CACHE &&
          console.warn(`[Cache] Error deleting entry: ${key} from ${cache.opts.namespace}`);
        continue;
      }
@@ -157,7 +115,7 @@ async function clearExpiredFromCache(cache) {
  }

  if (cleared > 0) {
-    debugMemoryCache &&
+    cacheConfig.DEBUG_MEMORY_CACHE &&
      console.log(
        `[Cache] Cleared ${cleared} entries older than ${ttl}ms from ${cache.opts.namespace}`,
      );
@@ -198,7 +156,7 @@ async function clearAllExpiredFromCache() {
  }
 }

-if (!isRedisEnabled && !isEnabled(CI)) {
+if (!cacheConfig.USE_REDIS && !cacheConfig.CI) {
  /** @type {Set<NodeJS.Timeout>} */
  const cleanupIntervals = new Set();

@@ -209,7 +167,7 @@ if (!isRedisEnabled && !isEnabled(CI)) {

  cleanupIntervals.add(cleanup);

-  if (debugMemoryCache) {
+  if (cacheConfig.DEBUG_MEMORY_CACHE) {
    const monitor = setInterval(() => {
      const ttlStores = getTTLStores();
      const memory = process.memoryUsage();
@@ -230,13 +188,13 @@ if (!isRedisEnabled && !isEnabled(CI)) {
  }

  const dispose = () => {
-    debugMemoryCache && console.log('[Cache] Cleaning up and shutting down...');
+    cacheConfig.DEBUG_MEMORY_CACHE && console.log('[Cache] Cleaning up and shutting down...');
    cleanupIntervals.forEach((interval) => clearInterval(interval));
    cleanupIntervals.clear();

    // One final cleanup before exit
    clearAllExpiredFromCache().then(() => {
-      debugMemoryCache && console.log('[Cache] Final cleanup completed');
+      cacheConfig.DEBUG_MEMORY_CACHE && console.log('[Cache] Final cleanup completed');
      process.exit(0);
    });
  };
--- a/api/cache/keyvFiles.js
+++ b/api/cache/keyvFiles.js
@@ -1,11 +1,9 @@
 const { KeyvFile } = require('keyv-file');

-const logFile = new KeyvFile({ filename: './data/logs.json' });
-const pendingReqFile = new KeyvFile({ filename: './data/pendingReqCache.json' });
-const violationFile = new KeyvFile({ filename: './data/violations.json' });
+const logFile = new KeyvFile({ filename: './data/logs.json' }).setMaxListeners(20);
+const violationFile = new KeyvFile({ filename: './data/violations.json' }).setMaxListeners(20);

 module.exports = {
  logFile,
-  pendingReqFile,
  violationFile,
 };
--- a/api/cache/keyvMongo.js
+++ b/api/cache/keyvMongo.js
@@ -1,9 +1,272 @@
-const KeyvMongo = require('@keyv/mongo');
+// api/cache/keyvMongo.js
+const mongoose = require('mongoose');
+const EventEmitter = require('events');
+const { GridFSBucket } = require('mongodb');
 const { logger } = require('~/config');

-const { MONGO_URI } = process.env ?? {};
+const storeMap = new Map();
+
+class KeyvMongoCustom extends EventEmitter {
+  constructor(url, options = {}) {
+    super();
+
+    url = url || {};
+    if (typeof url === 'string') {
+      url = { url };
+    }
+    if (url.uri) {
+      url = { url: url.uri, ...url };
+    }
+
+    this.opts = {
+      url: 'mongodb://127.0.0.1:27017',
+      collection: 'keyv',
+      ...url,
+      ...options,
+    };
+
+    this.ttlSupport = false;
+
+    // Filter valid options
+    const keyvMongoKeys = new Set([
+      'url',
+      'collection',
+      'namespace',
+      'serialize',
+      'deserialize',
+      'uri',
+      'useGridFS',
+      'dialect',
+    ]);
+    this.opts = Object.fromEntries(Object.entries(this.opts).filter(([k]) => keyvMongoKeys.has(k)));
+  }
+
+  // Helper to access the store WITHOUT storing a promise on the instance
+  _getClient() {
+    const storeKey = `${this.opts.collection}:${this.opts.useGridFS ? 'gridfs' : 'collection'}`;
+
+    // If we already have the store initialized, return it directly
+    if (storeMap.has(storeKey)) {
+      return Promise.resolve(storeMap.get(storeKey));
+    }
+
+    // Check mongoose connection state
+    if (mongoose.connection.readyState !== 1) {
+      return Promise.reject(
+        new Error('Mongoose connection not ready. Ensure connectDb() is called first.'),
+      );
+    }
+
+    try {
+      const db = mongoose.connection.db;
+      let client;
+
+      if (this.opts.useGridFS) {
+        const bucket = new GridFSBucket(db, {
+          readPreference: this.opts.readPreference,
+          bucketName: this.opts.collection,
+        });
+        const store = db.collection(`${this.opts.collection}.files`);
+        client = { bucket, store, db };
+      } else {
+        const collection = this.opts.collection || 'keyv';
+        const store = db.collection(collection);
+        client = { store, db };
+      }
+
+      storeMap.set(storeKey, client);
+      return Promise.resolve(client);
+    } catch (error) {
+      this.emit('error', error);
+      return Promise.reject(error);
+    }
+  }
+
+  async get(key) {
+    const client = await this._getClient();
+
+    if (this.opts.useGridFS) {
+      await client.store.updateOne(
+        {
+          filename: key,
+        },
+        {
+          $set: {
+            'metadata.lastAccessed': new Date(),
+          },
+        },
+      );
+
+      const stream = client.bucket.openDownloadStreamByName(key);
+
+      return new Promise((resolve) => {
+        const resp = [];
+        stream.on('error', () => {
+          resolve(undefined);
+        });
+
+        stream.on('end', () => {
+          const data = Buffer.concat(resp).toString('utf8');
+          resolve(data);
+        });
+
+        stream.on('data', (chunk) => {
+          resp.push(chunk);
+        });
+      });
+    }
+
+    const document = await client.store.findOne({ key: { $eq: key } });
+
+    if (!document) {
+      return undefined;
+    }
+
+    return document.value;
+  }
+
+  async getMany(keys) {
+    const client = await this._getClient();
+
+    if (this.opts.useGridFS) {
+      const promises = [];
+      for (const key of keys) {
+        promises.push(this.get(key));
+      }
+
+      const values = await Promise.allSettled(promises);
+      const data = [];
+      for (const value of values) {
+        data.push(value.value);
+      }
+
+      return data;
+    }
+
+    const values = await client.store
+      .find({ key: { $in: keys } })
+      .project({ _id: 0, value: 1, key: 1 })
+      .toArray();
+
+    const results = [...keys];
+    let i = 0;
+    for (const key of keys) {
+      const rowIndex = values.findIndex((row) => row.key === key);
+      results[i] = rowIndex > -1 ? values[rowIndex].value : undefined;
+      i++;
+    }
+
+    return results;
+  }
+
+  async set(key, value, ttl) {
+    const client = await this._getClient();
+    const expiresAt = typeof ttl === 'number' ? new Date(Date.now() + ttl) : null;
+
+    if (this.opts.useGridFS) {
+      const stream = client.bucket.openUploadStream(key, {
+        metadata: {
+          expiresAt,
+          lastAccessed: new Date(),
+        },
+      });
+
+      return new Promise((resolve) => {
+        stream.on('finish', () => {
+          resolve(stream);
+        });
+        stream.end(value);
+      });
+    }
+
+    await client.store.updateOne(
+      { key: { $eq: key } },
+      { $set: { key, value, expiresAt } },
+      { upsert: true },
+    );
+  }
+
+  async delete(key) {
+    if (typeof key !== 'string') {
+      return false;
+    }
+
+    const client = await this._getClient();
+
+    if (this.opts.useGridFS) {
+      try {
+        const bucket = new GridFSBucket(client.db, {
+          bucketName: this.opts.collection,
+        });
+        const files = await bucket.find({ filename: key }).toArray();
+        await client.bucket.delete(files[0]._id);
+        return true;
+      } catch {
+        return false;
+      }
+    }
+
+    const object = await client.store.deleteOne({ key: { $eq: key } });
+    return object.deletedCount > 0;
+  }
+
+  async deleteMany(keys) {
+    const client = await this._getClient();
+
+    if (this.opts.useGridFS) {
+      const bucket = new GridFSBucket(client.db, {
+        bucketName: this.opts.collection,
+      });
+      const files = await bucket.find({ filename: { $in: keys } }).toArray();
+      if (files.length === 0) {
+        return false;
+      }
+
+      await Promise.all(files.map(async (file) => client.bucket.delete(file._id)));
+      return true;
+    }
+
+    const object = await client.store.deleteMany({ key: { $in: keys } });
+    return object.deletedCount > 0;
+  }
+
+  async clear() {
+    const client = await this._getClient();
+
+    if (this.opts.useGridFS) {
+      try {
+        await client.bucket.drop();
+      } catch (error) {
+        // Throw error if not "namespace not found" error
+        if (!(error.code === 26)) {
+          throw error;
+        }
+      }
+    }
+
+    await client.store.deleteMany({
+      key: { $regex: this.namespace ? `^${this.namespace}:*` : '' },
+    });
+  }
+
+  async has(key) {
+    const client = await this._getClient();
+    const filter = { [this.opts.useGridFS ? 'filename' : 'key']: { $eq: key } };
+    const document = await client.store.countDocuments(filter, { limit: 1 });
+    return document !== 0;
+  }
+
+  // No-op disconnect
+  async disconnect() {
+    // This is a no-op since we don't want to close the shared mongoose connection
+    return true;
+  }
+}
+
+const keyvMongo = new KeyvMongoCustom({
+  collection: 'logs',
+});

-const keyvMongo = new KeyvMongo(MONGO_URI, { collection: 'logs' });
 keyvMongo.on('error', (err) => logger.error('KeyvMongo connection error:', err));

 module.exports = keyvMongo;
--- a/api/cache/keyvRedis.js
+++ b/api/cache/keyvRedis.js
@@ -1,86 +0,0 @@
-const fs = require('fs');
-const ioredis = require('ioredis');
-const KeyvRedis = require('@keyv/redis');
-const { isEnabled } = require('~/server/utils');
-const logger = require('~/config/winston');
-
-const { REDIS_URI, USE_REDIS, USE_REDIS_CLUSTER, REDIS_CA, REDIS_KEY_PREFIX, REDIS_MAX_LISTENERS } =
-  process.env;
-
-let keyvRedis;
-const redis_prefix = REDIS_KEY_PREFIX || '';
-const redis_max_listeners = REDIS_MAX_LISTENERS || 10;
-
-function mapURI(uri) {
-  const regex =
-    /^(?:(?<scheme>\w+):\/\/)?(?:(?<user>[^:@]+)(?::(?<password>[^@]+))?@)?(?<host>[\w.-]+)(?::(?<port>\d{1,5}))?$/;
-  const match = uri.match(regex);
-
-  if (match) {
-    const { scheme, user, password, host, port } = match.groups;
-
-    return {
-      scheme: scheme || 'none',
-      user: user || null,
-      password: password || null,
-      host: host || null,
-      port: port || null,
-    };
-  } else {
-    const parts = uri.split(':');
-    if (parts.length === 2) {
-      return {
-        scheme: 'none',
-        user: null,
-        password: null,
-        host: parts[0],
-        port: parts[1],
-      };
-    }
-
-    return {
-      scheme: 'none',
-      user: null,
-      password: null,
-      host: uri,
-      port: null,
-    };
-  }
-}
-
-if (REDIS_URI && isEnabled(USE_REDIS)) {
-  let redisOptions = null;
-  let keyvOpts = {
-    useRedisSets: false,
-    keyPrefix: redis_prefix,
-  };
-
-  if (REDIS_CA) {
-    const ca = fs.readFileSync(REDIS_CA);
-    redisOptions = { tls: { ca } };
-  }
-
-  if (isEnabled(USE_REDIS_CLUSTER)) {
-    const hosts = REDIS_URI.split(',').map((item) => {
-      var value = mapURI(item);
-
-      return {
-        host: value.host,
-        port: value.port,
-      };
-    });
-    const cluster = new ioredis.Cluster(hosts, { redisOptions });
-    keyvRedis = new KeyvRedis(cluster, keyvOpts);
-  } else {
-    keyvRedis = new KeyvRedis(REDIS_URI, keyvOpts);
-  }
-  keyvRedis.on('error', (err) => logger.error('KeyvRedis connection error:', err));
-  keyvRedis.setMaxListeners(redis_max_listeners);
-  logger.info(
-    '[Optional] Redis initialized. Note: Redis support is experimental. If you have issues, disable it. Cache needs to be flushed for values to refresh.',
-  );
-} else {
-  logger.info('[Optional] Redis not initialized. Note: Redis support is experimental.');
-}
-
-module.exports = keyvRedis;
--- a/api/cache/logViolation.js
+++ b/api/cache/logViolation.js
@@ -1,4 +1,5 @@
 const { isEnabled } = require('~/server/utils');
+const { ViolationTypes } = require('librechat-data-provider');
 const getLogStores = require('./getLogStores');
 const banViolation = require('./banViolation');

@@ -9,14 +10,14 @@ const banViolation = require('./banViolation');
 * @param {Object} res - Express response object.
 * @param {string} type - The type of violation.
 * @param {Object} errorMessage - The error message to log.
- * @param {number} [score=1] - The severity of the violation. Defaults to 1
+ * @param {number | string} [score=1] - The severity of the violation. Defaults to 1
 */
 const logViolation = async (req, res, type, errorMessage, score = 1) => {
  const userId = req.user?.id ?? req.user?._id;
  if (!userId) {
    return;
  }
-  const logs = getLogStores('general');
+  const logs = getLogStores(ViolationTypes.GENERAL);
  const violationLogs = getLogStores(type);
  const key = isEnabled(process.env.USE_REDIS) ? `${type}:${userId}` : userId;

--- a/api/cache/redis.js
+++ b/api/cache/redis.js
@@ -1,4 +0,0 @@
-const Redis = require('ioredis');
-const { REDIS_URI } = process.env ?? {};
-const redis = new Redis.Cluster(REDIS_URI);
-module.exports = redis;
--- a/api/cache/redisClients.js
+++ b/api/cache/redisClients.js
@@ -0,0 +1,96 @@
+const IoRedis = require('ioredis');
+const { logger } = require('@librechat/data-schemas');
+const { createClient, createCluster } = require('@keyv/redis');
+const { cacheConfig } = require('./cacheConfig');
+
+const GLOBAL_PREFIX_SEPARATOR = '::';
+
+const urls = cacheConfig.REDIS_URI?.split(',').map((uri) => new URL(uri));
+const username = urls?.[0].username || cacheConfig.REDIS_USERNAME;
+const password = urls?.[0].password || cacheConfig.REDIS_PASSWORD;
+const ca = cacheConfig.REDIS_CA;
+
+/** @type {import('ioredis').Redis | import('ioredis').Cluster | null} */
+let ioredisClient = null;
+if (cacheConfig.USE_REDIS) {
+  const redisOptions = {
+    username: username,
+    password: password,
+    tls: ca ? { ca } : undefined,
+    keyPrefix: `${cacheConfig.REDIS_KEY_PREFIX}${GLOBAL_PREFIX_SEPARATOR}`,
+    maxListeners: cacheConfig.REDIS_MAX_LISTENERS,
+  };
+
+  ioredisClient =
+    urls.length === 1
+      ? new IoRedis(cacheConfig.REDIS_URI, redisOptions)
+      : new IoRedis.Cluster(cacheConfig.REDIS_URI, { redisOptions });
+
+  ioredisClient.on('error', (err) => {
+    logger.error('ioredis client error:', err);
+  });
+
+  /** Ping Interval to keep the Redis server connection alive (if enabled) */
+  let pingInterval = null;
+  const clearPingInterval = () => {
+    if (pingInterval) {
+      clearInterval(pingInterval);
+      pingInterval = null;
+    }
+  };
+
+  if (cacheConfig.REDIS_PING_INTERVAL > 0) {
+    pingInterval = setInterval(() => {
+      if (ioredisClient && ioredisClient.status === 'ready') {
+        ioredisClient.ping();
+      }
+    }, cacheConfig.REDIS_PING_INTERVAL * 1000);
+    ioredisClient.on('close', clearPingInterval);
+    ioredisClient.on('end', clearPingInterval);
+  }
+}
+
+/** @type {import('@keyv/redis').RedisClient | import('@keyv/redis').RedisCluster | null} */
+let keyvRedisClient = null;
+if (cacheConfig.USE_REDIS) {
+  /**
+   * ** WARNING ** Keyv Redis client does not support Prefix like ioredis above.
+   * The prefix feature will be handled by the Keyv-Redis store in cacheFactory.js
+   */
+  const redisOptions = { username, password, socket: { tls: ca != null, ca } };
+
+  keyvRedisClient =
+    urls.length === 1
+      ? createClient({ url: cacheConfig.REDIS_URI, ...redisOptions })
+      : createCluster({
+          rootNodes: cacheConfig.REDIS_URI.split(',').map((url) => ({ url })),
+          defaults: redisOptions,
+        });
+
+  keyvRedisClient.setMaxListeners(cacheConfig.REDIS_MAX_LISTENERS);
+
+  keyvRedisClient.on('error', (err) => {
+    logger.error('@keyv/redis client error:', err);
+  });
+
+  /** Ping Interval to keep the Redis server connection alive (if enabled) */
+  let pingInterval = null;
+  const clearPingInterval = () => {
+    if (pingInterval) {
+      clearInterval(pingInterval);
+      pingInterval = null;
+    }
+  };
+
+  if (cacheConfig.REDIS_PING_INTERVAL > 0) {
+    pingInterval = setInterval(() => {
+      if (keyvRedisClient && keyvRedisClient.isReady) {
+        keyvRedisClient.ping();
+      }
+    }, cacheConfig.REDIS_PING_INTERVAL * 1000);
+    keyvRedisClient.on('disconnect', clearPingInterval);
+    keyvRedisClient.on('end', clearPingInterval);
+  }
+}
+
+module.exports = { ioredisClient, keyvRedisClient, GLOBAL_PREFIX_SEPARATOR };
--- a/api/config/index.js
+++ b/api/config/index.js
@@ -1,55 +1,42 @@
 const { EventSource } = require('eventsource');
-const { Time, CacheKeys } = require('librechat-data-provider');
+const { Time } = require('librechat-data-provider');
+const { MCPManager, FlowStateManager } = require('@librechat/api');
 const logger = require('./winston');

 global.EventSource = EventSource;

+/** @type {MCPManager} */
 let mcpManager = null;
 let flowManager = null;

 /**
- * @returns {Promise<MCPManager>}
+ * @param {string} [userId] - Optional user ID, to avoid disconnecting the current user.
+ * @returns {MCPManager}
 */
-async function getMCPManager() {
+function getMCPManager(userId) {
  if (!mcpManager) {
-    const { MCPManager } = await import('librechat-mcp');
-    mcpManager = MCPManager.getInstance(logger);
+    mcpManager = MCPManager.getInstance();
+  } else {
+    mcpManager.checkIdleConnections(userId);
  }
  return mcpManager;
 }

 /**
- * @param {(key: string) => Keyv} getLogStores
- * @returns {Promise<FlowStateManager>}
+ * @param {Keyv} flowsCache
+ * @returns {FlowStateManager}
 */
-async function getFlowStateManager(getLogStores) {
+function getFlowStateManager(flowsCache) {
  if (!flowManager) {
-    const { FlowStateManager } = await import('librechat-mcp');
-    flowManager = new FlowStateManager(getLogStores(CacheKeys.FLOWS), {
+    flowManager = new FlowStateManager(flowsCache, {
      ttl: Time.ONE_MINUTE * 3,
-      logger,
    });
  }
  return flowManager;
 }

-/**
- * Sends message data in Server Sent Events format.
- * @param {ServerResponse} res - The server response.
- * @param {{ data: string | Record<string, unknown>, event?: string }} event - The message event.
- * @param {string} event.event - The type of event.
- * @param {string} event.data - The message to be sent.
- */
-const sendEvent = (res, event) => {
-  if (typeof event.data === 'string' && event.data.length === 0) {
-    return;
-  }
-  res.write(`event: message\ndata: ${JSON.stringify(event)}\n\n`);
-};
-
 module.exports = {
  logger,
-  sendEvent,
  getMCPManager,
  getFlowStateManager,
 };
--- a/api/config/meiliLogger.js
+++ b/api/config/meiliLogger.js
@@ -4,7 +4,11 @@ require('winston-daily-rotate-file');

 const logDir = path.join(__dirname, '..', 'logs');

-const { NODE_ENV } = process.env;
+const { NODE_ENV, DEBUG_LOGGING = false } = process.env;
+
+const useDebugLogging =
+  (typeof DEBUG_LOGGING === 'string' && DEBUG_LOGGING?.toLowerCase() === 'true') ||
+  DEBUG_LOGGING === true;

 const levels = {
  error: 0,
@@ -36,9 +40,10 @@ const fileFormat = winston.format.combine(
  winston.format.splat(),
 );

+const logLevel = useDebugLogging ? 'debug' : 'error';
 const transports = [
  new winston.transports.DailyRotateFile({
-    level: 'debug',
+    level: logLevel,
    filename: `${logDir}/meiliSync-%DATE%.log`,
    datePattern: 'YYYY-MM-DD',
    zippedArchive: true,
@@ -48,14 +53,6 @@ const transports = [
  }),
 ];

-// if (NODE_ENV !== 'production') {
-//   transports.push(
-//     new winston.transports.Console({
-//       format: winston.format.combine(winston.format.colorize(), winston.format.simple()),
-//     }),
-//   );
-// }
-
 const consoleFormat = winston.format.combine(
  winston.format.colorize({ all: true }),
  winston.format.timestamp({ format: 'YYYY-MM-DD HH:mm:ss' }),
--- a/api/config/winston.js
+++ b/api/config/winston.js
@@ -5,7 +5,7 @@ const { redactFormat, redactMessage, debugTraverse, jsonTruncateFormat } = requi

 const logDir = path.join(__dirname, '..', 'logs');

-const { NODE_ENV, DEBUG_LOGGING = true, DEBUG_CONSOLE = false, CONSOLE_JSON = false } = process.env;
+const { NODE_ENV, DEBUG_LOGGING = true, CONSOLE_JSON = false, DEBUG_CONSOLE = false } = process.env;

 const useConsoleJson =
  (typeof CONSOLE_JSON === 'string' && CONSOLE_JSON?.toLowerCase() === 'true') ||
@@ -15,6 +15,10 @@ const useDebugConsole =
  (typeof DEBUG_CONSOLE === 'string' && DEBUG_CONSOLE?.toLowerCase() === 'true') ||
  DEBUG_CONSOLE === true;

+const useDebugLogging =
+  (typeof DEBUG_LOGGING === 'string' && DEBUG_LOGGING?.toLowerCase() === 'true') ||
+  DEBUG_LOGGING === true;
+
 const levels = {
  error: 0,
  warn: 1,
@@ -57,28 +61,9 @@ const transports = [
    maxFiles: '14d',
    format: fileFormat,
  }),
-  // new winston.transports.DailyRotateFile({
-  //   level: 'info',
-  //   filename: `${logDir}/info-%DATE%.log`,
-  //   datePattern: 'YYYY-MM-DD',
-  //   zippedArchive: true,
-  //   maxSize: '20m',
-  //   maxFiles: '14d',
-  // }),
 ];

-// if (NODE_ENV !== 'production') {
-//   transports.push(
-//     new winston.transports.Console({
-//       format: winston.format.combine(winston.format.colorize(), winston.format.simple()),
-//     }),
-//   );
-// }
-
-if (
-  (typeof DEBUG_LOGGING === 'string' && DEBUG_LOGGING?.toLowerCase() === 'true') ||
-  DEBUG_LOGGING === true
-) {
+if (useDebugLogging) {
  transports.push(
    new winston.transports.DailyRotateFile({
      level: 'debug',
@@ -107,10 +92,16 @@ const consoleFormat = winston.format.combine(
  }),
 );

+// Determine console log level
+let consoleLogLevel = 'info';
+if (useDebugConsole) {
+  consoleLogLevel = 'debug';
+}
+
 if (useDebugConsole) {
  transports.push(
    new winston.transports.Console({
-      level: 'debug',
+      level: consoleLogLevel,
      format: useConsoleJson
        ? winston.format.combine(fileFormat, jsonTruncateFormat(), winston.format.json())
        : winston.format.combine(fileFormat, debugTraverse),
@@ -119,14 +110,14 @@ if (useDebugConsole) {
 } else if (useConsoleJson) {
  transports.push(
    new winston.transports.Console({
-      level: 'info',
+      level: consoleLogLevel,
      format: winston.format.combine(fileFormat, jsonTruncateFormat(), winston.format.json()),
    }),
  );
 } else {
  transports.push(
    new winston.transports.Console({
-      level: 'info',
+      level: consoleLogLevel,
      format: consoleFormat,
    }),
  );
--- a/api/lib/db/connectDb.js
+++ b/api/lib/db/connectDb.js
@@ -39,7 +39,10 @@ async function connectDb() {
    });
  }
  cached.conn = await cached.promise;
+
  return cached.conn;
 }

-module.exports = connectDb;
+module.exports = {
+  connectDb,
+};
--- a/api/db/index.js
+++ b/api/db/index.js
@@ -0,0 +1,8 @@
+const mongoose = require('mongoose');
+const { createModels } = require('@librechat/data-schemas');
+const { connectDb } = require('./connect');
+const indexSync = require('./indexSync');
+
+createModels(mongoose);
+
+module.exports = { connectDb, indexSync };
--- a/api/db/indexSync.js
+++ b/api/db/indexSync.js
@@ -0,0 +1,174 @@
+const mongoose = require('mongoose');
+const { MeiliSearch } = require('meilisearch');
+const { logger } = require('@librechat/data-schemas');
+const { FlowStateManager } = require('@librechat/api');
+const { CacheKeys } = require('librechat-data-provider');
+
+const { isEnabled } = require('~/server/utils');
+const { getLogStores } = require('~/cache');
+
+const Conversation = mongoose.models.Conversation;
+const Message = mongoose.models.Message;
+
+const searchEnabled = isEnabled(process.env.SEARCH);
+const indexingDisabled = isEnabled(process.env.MEILI_NO_SYNC);
+let currentTimeout = null;
+
+class MeiliSearchClient {
+  static instance = null;
+
+  static getInstance() {
+    if (!MeiliSearchClient.instance) {
+      if (!process.env.MEILI_HOST || !process.env.MEILI_MASTER_KEY) {
+        throw new Error('Meilisearch configuration is missing.');
+      }
+      MeiliSearchClient.instance = new MeiliSearch({
+        host: process.env.MEILI_HOST,
+        apiKey: process.env.MEILI_MASTER_KEY,
+      });
+    }
+    return MeiliSearchClient.instance;
+  }
+}
+
+/**
+ * Performs the actual sync operations for messages and conversations
+ */
+async function performSync() {
+  const client = MeiliSearchClient.getInstance();
+
+  const { status } = await client.health();
+  if (status !== 'available') {
+    throw new Error('Meilisearch not available');
+  }
+
+  if (indexingDisabled === true) {
+    logger.info('[indexSync] Indexing is disabled, skipping...');
+    return { messagesSync: false, convosSync: false };
+  }
+
+  let messagesSync = false;
+  let convosSync = false;
+
+  // Check if we need to sync messages
+  const messageProgress = await Message.getSyncProgress();
+  if (!messageProgress.isComplete) {
+    logger.info(
+      `[indexSync] Messages need syncing: ${messageProgress.totalProcessed}/${messageProgress.totalDocuments} indexed`,
+    );
+
+    // Check if we should do a full sync or incremental
+    const messageCount = await Message.countDocuments();
+    const messagesIndexed = messageProgress.totalProcessed;
+    const syncThreshold = parseInt(process.env.MEILI_SYNC_THRESHOLD || '1000', 10);
+
+    if (messageCount - messagesIndexed > syncThreshold) {
+      logger.info('[indexSync] Starting full message sync due to large difference');
+      await Message.syncWithMeili();
+      messagesSync = true;
+    } else if (messageCount !== messagesIndexed) {
+      logger.warn('[indexSync] Messages out of sync, performing incremental sync');
+      await Message.syncWithMeili();
+      messagesSync = true;
+    }
+  } else {
+    logger.info(
+      `[indexSync] Messages are fully synced: ${messageProgress.totalProcessed}/${messageProgress.totalDocuments}`,
+    );
+  }
+
+  // Check if we need to sync conversations
+  const convoProgress = await Conversation.getSyncProgress();
+  if (!convoProgress.isComplete) {
+    logger.info(
+      `[indexSync] Conversations need syncing: ${convoProgress.totalProcessed}/${convoProgress.totalDocuments} indexed`,
+    );
+
+    const convoCount = await Conversation.countDocuments();
+    const convosIndexed = convoProgress.totalProcessed;
+    const syncThreshold = parseInt(process.env.MEILI_SYNC_THRESHOLD || '1000', 10);
+
+    if (convoCount - convosIndexed > syncThreshold) {
+      logger.info('[indexSync] Starting full conversation sync due to large difference');
+      await Conversation.syncWithMeili();
+      convosSync = true;
+    } else if (convoCount !== convosIndexed) {
+      logger.warn('[indexSync] Convos out of sync, performing incremental sync');
+      await Conversation.syncWithMeili();
+      convosSync = true;
+    }
+  } else {
+    logger.info(
+      `[indexSync] Conversations are fully synced: ${convoProgress.totalProcessed}/${convoProgress.totalDocuments}`,
+    );
+  }
+
+  return { messagesSync, convosSync };
+}
+
+/**
+ * Main index sync function that uses FlowStateManager to prevent concurrent execution
+ */
+async function indexSync() {
+  if (!searchEnabled) {
+    return;
+  }
+
+  logger.info('[indexSync] Starting index synchronization check...');
+
+  try {
+    // Get or create FlowStateManager instance
+    const flowsCache = getLogStores(CacheKeys.FLOWS);
+    if (!flowsCache) {
+      logger.warn('[indexSync] Flows cache not available, falling back to direct sync');
+      return await performSync();
+    }
+
+    const flowManager = new FlowStateManager(flowsCache, {
+      ttl: 60000 * 10, // 10 minutes TTL for sync operations
+    });
+
+    // Use a unique flow ID for the sync operation
+    const flowId = 'meili-index-sync';
+    const flowType = 'MEILI_SYNC';
+
+    // This will only execute the handler if no other instance is running the sync
+    const result = await flowManager.createFlowWithHandler(flowId, flowType, performSync);
+
+    if (result.messagesSync || result.convosSync) {
+      logger.info('[indexSync] Sync completed successfully');
+    } else {
+      logger.debug('[indexSync] No sync was needed');
+    }
+
+    return result;
+  } catch (err) {
+    if (err.message.includes('flow already exists')) {
+      logger.info('[indexSync] Sync already running on another instance');
+      return;
+    }
+
+    if (err.message.includes('not found')) {
+      logger.debug('[indexSync] Creating indices...');
+      currentTimeout = setTimeout(async () => {
+        try {
+          await Message.syncWithMeili();
+          await Conversation.syncWithMeili();
+        } catch (err) {
+          logger.error('[indexSync] Trouble creating indices, try restarting the server.', err);
+        }
+      }, 750);
+    } else if (err.message.includes('Meilisearch not configured')) {
+      logger.info('[indexSync] Meilisearch not configured, search will be disabled.');
+    } else {
+      logger.error('[indexSync] error', err);
+    }
+  }
+}
+
+process.on('exit', () => {
+  logger.debug('[indexSync] Clearing sync timeouts before exiting...');
+  clearTimeout(currentTimeout);
+});
+
+module.exports = indexSync;
--- a/api/db/models.js
+++ b/api/db/models.js
@@ -0,0 +1,5 @@
+const mongoose = require('mongoose');
+const { createModels } = require('@librechat/data-schemas');
+const models = createModels(mongoose);
+
+module.exports = { ...models };
--- a/api/jest.config.js
+++ b/api/jest.config.js
@@ -5,12 +5,14 @@ module.exports = {
  coverageDirectory: 'coverage',
  setupFiles: [
    './test/jestSetup.js',
-    './test/__mocks__/KeyvMongo.js',
    './test/__mocks__/logger.js',
    './test/__mocks__/fetchEventSource.js',
  ],
  moduleNameMapper: {
    '~/(.*)': '<rootDir>/$1',
    '~/data/auth.json': '<rootDir>/__mocks__/auth.mock.json',
+    '^openid-client/passport$': '<rootDir>/test/__mocks__/openid-client-passport.js', // Mock for the passport strategy part
+    '^openid-client$': '<rootDir>/test/__mocks__/openid-client.js',
  },
+  transformIgnorePatterns: ['/node_modules/(?!(openid-client|oauth4webapi|jose)/).*/'],
 };
--- a/api/lib/db/index.js
+++ b/api/lib/db/index.js
@@ -1,4 +0,0 @@
-const connectDb = require('./connectDb');
-const indexSync = require('./indexSync');
-
-module.exports = { connectDb, indexSync };
--- a/api/lib/db/indexSync.js
+++ b/api/lib/db/indexSync.js
@@ -1,89 +0,0 @@
-const { MeiliSearch } = require('meilisearch');
-const Conversation = require('~/models/schema/convoSchema');
-const Message = require('~/models/schema/messageSchema');
-const { isEnabled } = require('~/server/utils');
-const { logger } = require('~/config');
-
-const searchEnabled = isEnabled(process.env.SEARCH);
-const indexingDisabled = isEnabled(process.env.MEILI_NO_SYNC);
-let currentTimeout = null;
-
-class MeiliSearchClient {
-  static instance = null;
-
-  static getInstance() {
-    if (!MeiliSearchClient.instance) {
-      if (!process.env.MEILI_HOST || !process.env.MEILI_MASTER_KEY) {
-        throw new Error('Meilisearch configuration is missing.');
-      }
-      MeiliSearchClient.instance = new MeiliSearch({
-        host: process.env.MEILI_HOST,
-        apiKey: process.env.MEILI_MASTER_KEY,
-      });
-    }
-    return MeiliSearchClient.instance;
-  }
-}
-
-async function indexSync() {
-  if (!searchEnabled) {
-    return;
-  }
-
-  try {
-    const client = MeiliSearchClient.getInstance();
-
-    const { status } = await client.health();
-    if (status !== 'available') {
-      throw new Error('Meilisearch not available');
-    }
-
-    if (indexingDisabled === true) {
-      logger.info('[indexSync] Indexing is disabled, skipping...');
-      return;
-    }
-
-    const messageCount = await Message.countDocuments();
-    const convoCount = await Conversation.countDocuments();
-    const messages = await client.index('messages').getStats();
-    const convos = await client.index('convos').getStats();
-    const messagesIndexed = messages.numberOfDocuments;
-    const convosIndexed = convos.numberOfDocuments;
-
-    logger.debug(`[indexSync] There are ${messageCount} messages and ${messagesIndexed} indexed`);
-    logger.debug(`[indexSync] There are ${convoCount} convos and ${convosIndexed} indexed`);
-
-    if (messageCount !== messagesIndexed) {
-      logger.debug('[indexSync] Messages out of sync, indexing');
-      Message.syncWithMeili();
-    }
-
-    if (convoCount !== convosIndexed) {
-      logger.debug('[indexSync] Convos out of sync, indexing');
-      Conversation.syncWithMeili();
-    }
-  } catch (err) {
-    if (err.message.includes('not found')) {
-      logger.debug('[indexSync] Creating indices...');
-      currentTimeout = setTimeout(async () => {
-        try {
-          await Message.syncWithMeili();
-          await Conversation.syncWithMeili();
-        } catch (err) {
-          logger.error('[indexSync] Trouble creating indices, try restarting the server.', err);
-        }
-      }, 750);
-    } else if (err.message.includes('Meilisearch not configured')) {
-      logger.info('[indexSync] Meilisearch not configured, search will be disabled.');
-    } else {
-      logger.error('[indexSync] error', err);
-    }
-  }
-}
-
-process.on('exit', () => {
-  logger.debug('[indexSync] Clearing sync timeouts before exiting...');
-  clearTimeout(currentTimeout);
-});
-
-module.exports = indexSync;
--- a/api/lib/utils/reduceHits.js
+++ b/api/lib/utils/reduceHits.js
@@ -1,59 +0,0 @@
-const mergeSort = require('./mergeSort');
-const { cleanUpPrimaryKeyValue } = require('./misc');
-
-function reduceMessages(hits) {
-  const counts = {};
-
-  for (const hit of hits) {
-    if (!counts[hit.conversationId]) {
-      counts[hit.conversationId] = 1;
-    } else {
-      counts[hit.conversationId]++;
-    }
-  }
-
-  const result = [];
-
-  for (const [conversationId, count] of Object.entries(counts)) {
-    result.push({
-      conversationId,
-      count,
-    });
-  }
-
-  return mergeSort(result, (a, b) => b.count - a.count);
-}
-
-function reduceHits(hits, titles = []) {
-  const counts = {};
-  const titleMap = {};
-  const convos = [...hits, ...titles];
-
-  for (const convo of convos) {
-    const currentId = cleanUpPrimaryKeyValue(convo.conversationId);
-    if (!counts[currentId]) {
-      counts[currentId] = 1;
-    } else {
-      counts[currentId]++;
-    }
-
-    if (convo.title) {
-      // titleMap[currentId] = convo._formatted.title;
-      titleMap[currentId] = convo.title;
-    }
-  }
-
-  const result = [];
-
-  for (const [conversationId, count] of Object.entries(counts)) {
-    result.push({
-      conversationId,
-      count,
-      title: titleMap[conversationId] ? titleMap[conversationId] : null,
-    });
-  }
-
-  return mergeSort(result, (a, b) => b.count - a.count);
-}
-
-module.exports = { reduceMessages, reduceHits };
--- a/api/models/Action.js
+++ b/api/models/Action.js
@@ -1,7 +1,4 @@
-const mongoose = require('mongoose');
-const actionSchema = require('./schema/action');
-
-const Action = mongoose.model('action', actionSchema);
+const { Action } = require('~/db/models');

 /**
 * Update an action with new data without overwriting existing properties,
--- a/api/models/Agent.js
+++ b/api/models/Agent.js
@@ -1,6 +1,9 @@
 const mongoose = require('mongoose');
-const { SystemRoles } = require('librechat-data-provider');
-const { GLOBAL_PROJECT_NAME } = require('librechat-data-provider').Constants;
+const crypto = require('node:crypto');
+const { logger } = require('@librechat/data-schemas');
+const { SystemRoles, Tools, actionDelimiter } = require('librechat-data-provider');
+const { GLOBAL_PROJECT_NAME, EPHEMERAL_AGENT_ID, mcp_delimiter } =
+  require('librechat-data-provider').Constants;
 const { CONFIG_STORE, STARTUP_CONFIG } = require('librechat-data-provider').CacheKeys;
 const {
  getProjectByName,
@@ -8,10 +11,10 @@ const {
  removeAgentIdsFromProject,
  removeAgentFromAllProjects,
 } = require('./Project');
+const { getCachedTools } = require('~/server/services/Config');
 const getLogStores = require('~/cache/getLogStores');
-const agentSchema = require('./schema/agent');
-
-const Agent = mongoose.model('agent', agentSchema);
+const { getActions } = require('./Action');
+const { Agent } = require('~/db/models');

 /**
 * Create an agent with the provided data.
@@ -20,7 +23,19 @@ const Agent = mongoose.model('agent', agentSchema);
 * @throws {Error} If the agent creation fails.
 */
 const createAgent = async (agentData) => {
-  return (await Agent.create(agentData)).toObject();
+  const { author, ...versionData } = agentData;
+  const timestamp = new Date();
+  const initialAgentData = {
+    ...agentData,
+    versions: [
+      {
+        ...versionData,
+        createdAt: timestamp,
+        updatedAt: timestamp,
+      },
+    ],
+  };
+  return (await Agent.create(initialAgentData)).toObject();
 };

 /**
@@ -39,13 +54,84 @@ const getAgent = async (searchParameter) => await Agent.findOne(searchParameter)
 * @param {Object} params
 * @param {ServerRequest} params.req
 * @param {string} params.agent_id
+ * @param {string} params.endpoint
+ * @param {import('@librechat/agents').ClientOptions} [params.model_parameters]
 * @returns {Promise<Agent|null>} The agent document as a plain object, or null if not found.
 */
-const loadAgent = async ({ req, agent_id }) => {
+const loadEphemeralAgent = async ({ req, agent_id, endpoint, model_parameters: _m }) => {
+  const { model, ...model_parameters } = _m;
+  /** @type {Record<string, FunctionTool>} */
+  const availableTools = await getCachedTools({ userId: req.user.id, includeGlobal: true });
+  /** @type {TEphemeralAgent | null} */
+  const ephemeralAgent = req.body.ephemeralAgent;
+  const mcpServers = new Set(ephemeralAgent?.mcp);
+  /** @type {string[]} */
+  const tools = [];
+  if (ephemeralAgent?.execute_code === true) {
+    tools.push(Tools.execute_code);
+  }
+  if (ephemeralAgent?.file_search === true) {
+    tools.push(Tools.file_search);
+  }
+  if (ephemeralAgent?.web_search === true) {
+    tools.push(Tools.web_search);
+  }
+
+  if (mcpServers.size > 0) {
+    for (const toolName of Object.keys(availableTools)) {
+      if (!toolName.includes(mcp_delimiter)) {
+        continue;
+      }
+      const mcpServer = toolName.split(mcp_delimiter)?.[1];
+      if (mcpServer && mcpServers.has(mcpServer)) {
+        tools.push(toolName);
+      }
+    }
+  }
+
+  const instructions = req.body.promptPrefix;
+  const result = {
+    id: agent_id,
+    instructions,
+    provider: endpoint,
+    model_parameters,
+    model,
+    tools,
+  };
+
+  if (ephemeralAgent?.artifacts != null && ephemeralAgent.artifacts) {
+    result.artifacts = ephemeralAgent.artifacts;
+  }
+  return result;
+};
+
+/**
+ * Load an agent based on the provided ID
+ *
+ * @param {Object} params
+ * @param {ServerRequest} params.req
+ * @param {string} params.agent_id
+ * @param {string} params.endpoint
+ * @param {import('@librechat/agents').ClientOptions} [params.model_parameters]
+ * @returns {Promise<Agent|null>} The agent document as a plain object, or null if not found.
+ */
+const loadAgent = async ({ req, agent_id, endpoint, model_parameters }) => {
+  if (!agent_id) {
+    return null;
+  }
+  if (agent_id === EPHEMERAL_AGENT_ID) {
+    return await loadEphemeralAgent({ req, agent_id, endpoint, model_parameters });
+  }
  const agent = await getAgent({
    id: agent_id,
  });

+  if (!agent) {
+    return null;
+  }
+
+  agent.version = agent.versions ? agent.versions.length : 0;
+
  if (agent.author.toString() === req.user.id) {
    return agent;
  }
@@ -70,19 +156,205 @@ const loadAgent = async ({ req, agent_id }) => {
  }
 };

+/**
+ * Check if a version already exists in the versions array, excluding timestamp and author fields
+ * @param {Object} updateData - The update data to compare
+ * @param {Object} currentData - The current agent data
+ * @param {Array} versions - The existing versions array
+ * @param {string} [actionsHash] - Hash of current action metadata
+ * @returns {Object|null} - The matching version if found, null otherwise
+ */
+const isDuplicateVersion = (updateData, currentData, versions, actionsHash = null) => {
+  if (!versions || versions.length === 0) {
+    return null;
+  }
+
+  const excludeFields = [
+    '_id',
+    'id',
+    'createdAt',
+    'updatedAt',
+    'author',
+    'updatedBy',
+    'created_at',
+    'updated_at',
+    '__v',
+    'versions',
+    'actionsHash', // Exclude actionsHash from direct comparison
+  ];
+
+  const { $push, $pull, $addToSet, ...directUpdates } = updateData;
+
+  if (Object.keys(directUpdates).length === 0 && !actionsHash) {
+    return null;
+  }
+
+  const wouldBeVersion = { ...currentData, ...directUpdates };
+  const lastVersion = versions[versions.length - 1];
+
+  if (actionsHash && lastVersion.actionsHash !== actionsHash) {
+    return null;
+  }
+
+  const allFields = new Set([...Object.keys(wouldBeVersion), ...Object.keys(lastVersion)]);
+
+  const importantFields = Array.from(allFields).filter((field) => !excludeFields.includes(field));
+
+  let isMatch = true;
+  for (const field of importantFields) {
+    if (!wouldBeVersion[field] && !lastVersion[field]) {
+      continue;
+    }
+
+    if (Array.isArray(wouldBeVersion[field]) && Array.isArray(lastVersion[field])) {
+      if (wouldBeVersion[field].length !== lastVersion[field].length) {
+        isMatch = false;
+        break;
+      }
+
+      // Special handling for projectIds (MongoDB ObjectIds)
+      if (field === 'projectIds') {
+        const wouldBeIds = wouldBeVersion[field].map((id) => id.toString()).sort();
+        const versionIds = lastVersion[field].map((id) => id.toString()).sort();
+
+        if (!wouldBeIds.every((id, i) => id === versionIds[i])) {
+          isMatch = false;
+          break;
+        }
+      }
+      // Handle arrays of objects like tool_kwargs
+      else if (typeof wouldBeVersion[field][0] === 'object' && wouldBeVersion[field][0] !== null) {
+        const sortedWouldBe = [...wouldBeVersion[field]].map((item) => JSON.stringify(item)).sort();
+        const sortedVersion = [...lastVersion[field]].map((item) => JSON.stringify(item)).sort();
+
+        if (!sortedWouldBe.every((item, i) => item === sortedVersion[i])) {
+          isMatch = false;
+          break;
+        }
+      } else {
+        const sortedWouldBe = [...wouldBeVersion[field]].sort();
+        const sortedVersion = [...lastVersion[field]].sort();
+
+        if (!sortedWouldBe.every((item, i) => item === sortedVersion[i])) {
+          isMatch = false;
+          break;
+        }
+      }
+    } else if (field === 'model_parameters') {
+      const wouldBeParams = wouldBeVersion[field] || {};
+      const lastVersionParams = lastVersion[field] || {};
+      if (JSON.stringify(wouldBeParams) !== JSON.stringify(lastVersionParams)) {
+        isMatch = false;
+        break;
+      }
+    } else if (wouldBeVersion[field] !== lastVersion[field]) {
+      isMatch = false;
+      break;
+    }
+  }
+
+  return isMatch ? lastVersion : null;
+};
+
 /**
 * Update an agent with new data without overwriting existing
 *  properties, or create a new agent if it doesn't exist.
+ * When an agent is updated, a copy of the current state will be saved to the versions array.
 *
 * @param {Object} searchParameter - The search parameters to find the agent to update.
 * @param {string} searchParameter.id - The ID of the agent to update.
 * @param {string} [searchParameter.author] - The user ID of the agent's author.
 * @param {Object} updateData - An object containing the properties to update.
+ * @param {Object} [options] - Optional configuration object.
+ * @param {string} [options.updatingUserId] - The ID of the user performing the update (used for tracking non-author updates).
+ * @param {boolean} [options.forceVersion] - Force creation of a new version even if no fields changed.
+ * @param {boolean} [options.skipVersioning] - Skip version creation entirely (useful for isolated operations like sharing).
 * @returns {Promise<Agent>} The updated or newly created agent document as a plain object.
+ * @throws {Error} If the update would create a duplicate version
 */
-const updateAgent = async (searchParameter, updateData) => {
-  const options = { new: true, upsert: false };
-  return Agent.findOneAndUpdate(searchParameter, updateData, options).lean();
+const updateAgent = async (searchParameter, updateData, options = {}) => {
+  const { updatingUserId = null, forceVersion = false, skipVersioning = false } = options;
+  const mongoOptions = { new: true, upsert: false };
+
+  const currentAgent = await Agent.findOne(searchParameter);
+  if (currentAgent) {
+    const { __v, _id, id, versions, author, ...versionData } = currentAgent.toObject();
+    const { $push, $pull, $addToSet, ...directUpdates } = updateData;
+
+    let actionsHash = null;
+
+    // Generate actions hash if agent has actions
+    if (currentAgent.actions && currentAgent.actions.length > 0) {
+      // Extract action IDs from the format "domain_action_id"
+      const actionIds = currentAgent.actions
+        .map((action) => {
+          const parts = action.split(actionDelimiter);
+          return parts[1]; // Get just the action ID part
+        })
+        .filter(Boolean);
+
+      if (actionIds.length > 0) {
+        try {
+          const actions = await getActions(
+            {
+              action_id: { $in: actionIds },
+            },
+            true,
+          ); // Include sensitive data for hash
+
+          actionsHash = await generateActionMetadataHash(currentAgent.actions, actions);
+        } catch (error) {
+          logger.error('Error fetching actions for hash generation:', error);
+        }
+      }
+    }
+
+    const shouldCreateVersion =
+      !skipVersioning &&
+      (forceVersion || Object.keys(directUpdates).length > 0 || $push || $pull || $addToSet);
+
+    if (shouldCreateVersion) {
+      const duplicateVersion = isDuplicateVersion(updateData, versionData, versions, actionsHash);
+      if (duplicateVersion && !forceVersion) {
+        const error = new Error(
+          'Duplicate version: This would create a version identical to an existing one',
+        );
+        error.statusCode = 409;
+        error.details = {
+          duplicateVersion,
+          versionIndex: versions.findIndex(
+            (v) => JSON.stringify(duplicateVersion) === JSON.stringify(v),
+          ),
+        };
+        throw error;
+      }
+    }
+
+    const versionEntry = {
+      ...versionData,
+      ...directUpdates,
+      updatedAt: new Date(),
+    };
+
+    // Include actions hash in version if available
+    if (actionsHash) {
+      versionEntry.actionsHash = actionsHash;
+    }
+
+    // Always store updatedBy field to track who made the change
+    if (updatingUserId) {
+      versionEntry.updatedBy = new mongoose.Types.ObjectId(updatingUserId);
+    }
+
+    if (shouldCreateVersion) {
+      updateData.$push = {
+        ...($push || {}),
+        versions: versionEntry,
+      };
+    }
+  }
+
+  return Agent.findOneAndUpdate(searchParameter, updateData, mongoOptions).lean();
 };

 /**
@@ -94,11 +366,13 @@ const updateAgent = async (searchParameter, updateData) => {
 * @param {string} params.file_id
 * @returns {Promise<Agent>} The updated agent.
 */
-const addAgentResourceFile = async ({ agent_id, tool_resource, file_id }) => {
+const addAgentResourceFile = async ({ req, agent_id, tool_resource, file_id }) => {
  const searchParameter = { id: agent_id };
-
+  let agent = await getAgent(searchParameter);
+  if (!agent) {
+    throw new Error('Agent not found for adding resource file');
+  }
  const fileIdsPath = `tool_resources.${tool_resource}.file_ids`;
-
  await Agent.updateOne(
    {
      id: agent_id,
@@ -111,9 +385,16 @@ const addAgentResourceFile = async ({ agent_id, tool_resource, file_id }) => {
    },
  );

-  const updateData = { $addToSet: { [fileIdsPath]: file_id } };
+  const updateData = {
+    $addToSet: {
+      tools: tool_resource,
+      [fileIdsPath]: file_id,
+    },
+  };

-  const updatedAgent = await updateAgent(searchParameter, updateData);
+  const updatedAgent = await updateAgent(searchParameter, updateData, {
+    updatingUserId: req?.user?.id,
+  });
  if (updatedAgent) {
    return updatedAgent;
  } else {
@@ -122,16 +403,17 @@ const addAgentResourceFile = async ({ agent_id, tool_resource, file_id }) => {
 };

 /**
- * Removes multiple resource files from an agent in a single update.
+ * Removes multiple resource files from an agent using atomic operations.
 * @param {object} params
 * @param {string} params.agent_id
 * @param {Array<{tool_resource: string, file_id: string}>} params.files
 * @returns {Promise<Agent>} The updated agent.
+ * @throws {Error} If the agent is not found or update fails.
 */
 const removeAgentResourceFiles = async ({ agent_id, files }) => {
  const searchParameter = { id: agent_id };

-  // associate each tool resource with the respective file ids array
+  // Group files to remove by resource
  const filesByResource = files.reduce((acc, { tool_resource, file_id }) => {
    if (!acc[tool_resource]) {
      acc[tool_resource] = [];
@@ -140,42 +422,35 @@ const removeAgentResourceFiles = async ({ agent_id, files }) => {
    return acc;
  }, {});

-  // build the update aggregation pipeline wich removes file ids from tool resources array
-  // and eventually deletes empty tool resources
-  const updateData = [];
-  Object.entries(filesByResource).forEach(([resource, fileIds]) => {
-    const toolResourcePath = `tool_resources.${resource}`;
-    const fileIdsPath = `${toolResourcePath}.file_ids`;
-
-    // file ids removal stage
-    updateData.push({
-      $set: {
-        [fileIdsPath]: {
-          $filter: {
-            input: `$${fileIdsPath}`,
-            cond: { $not: [{ $in: ['$$this', fileIds] }] },
-          },
-        },
-      },
-    });
-
-    // empty tool resource deletion stage
-    updateData.push({
-      $set: {
-        [toolResourcePath]: {
-          $cond: [{ $eq: [`$${fileIdsPath}`, []] }, '$$REMOVE', `$${toolResourcePath}`],
-        },
-      },
-    });
-  });
-
-  // return the updated agent or throw if no agent matches
-  const updatedAgent = await updateAgent(searchParameter, updateData);
-  if (updatedAgent) {
-    return updatedAgent;
-  } else {
-    throw new Error('Agent not found for removing resource files');
+  // Step 1: Atomically remove file IDs using $pull
+  const pullOps = {};
+  const resourcesToCheck = new Set();
+  for (const [resource, fileIds] of Object.entries(filesByResource)) {
+    const fileIdsPath = `tool_resources.${resource}.file_ids`;
+    pullOps[fileIdsPath] = { $in: fileIds };
+    resourcesToCheck.add(resource);
  }
+
+  const updatePullData = { $pull: pullOps };
+  const agentAfterPull = await Agent.findOneAndUpdate(searchParameter, updatePullData, {
+    new: true,
+  }).lean();
+
+  if (!agentAfterPull) {
+    // Agent might have been deleted concurrently, or never existed.
+    // Check if it existed before trying to throw.
+    const agentExists = await getAgent(searchParameter);
+    if (!agentExists) {
+      throw new Error('Agent not found for removing resource files');
+    }
+    // If it existed but findOneAndUpdate returned null, something else went wrong.
+    throw new Error('Failed to update agent during file removal (pull step)');
+  }
+
+  // Return the agent state directly after the $pull operation.
+  // Skipping the $unset step for now to simplify and test core $pull atomicity.
+  // Empty arrays might remain, but the removal itself should be correct.
+  return agentAfterPull;
 };

 /**
@@ -211,7 +486,6 @@ const getListAgents = async (searchParameter) => {
    delete globalQuery.author;
    query = { $or: [globalQuery, query] };
  }
-
  const agents = (
    await Agent.find(query, {
      id: 1,
@@ -250,7 +524,7 @@ const getListAgents = async (searchParameter) => {
 * This function also updates the corresponding projects to include or exclude the agent ID.
 *
 * @param {Object} params - Parameters for updating the agent's projects.
- * @param {import('librechat-data-provider').TUser} params.user - Parameters for updating the agent's projects.
+ * @param {MongoUser} params.user - Parameters for updating the agent's projects.
 * @param {string} params.agentId - The ID of the agent to update.
 * @param {string[]} [params.projectIds] - Array of project IDs to add to the agent.
 * @param {string[]} [params.removeProjectIds] - Array of project IDs to remove from the agent.
@@ -283,7 +557,10 @@ const updateAgentProjects = async ({ user, agentId, projectIds, removeProjectIds
    delete updateQuery.author;
  }

-  const updatedAgent = await updateAgent(updateQuery, updateOps);
+  const updatedAgent = await updateAgent(updateQuery, updateOps, {
+    updatingUserId: user.id,
+    skipVersioning: true,
+  });
  if (updatedAgent) {
    return updatedAgent;
  }
@@ -300,15 +577,107 @@ const updateAgentProjects = async ({ user, agentId, projectIds, removeProjectIds
  return await getAgent({ id: agentId });
 };

+/**
+ * Reverts an agent to a specific version in its version history.
+ * @param {Object} searchParameter - The search parameters to find the agent to revert.
+ * @param {string} searchParameter.id - The ID of the agent to revert.
+ * @param {string} [searchParameter.author] - The user ID of the agent's author.
+ * @param {number} versionIndex - The index of the version to revert to in the versions array.
+ * @returns {Promise<MongoAgent>} The updated agent document after reverting.
+ * @throws {Error} If the agent is not found or the specified version does not exist.
+ */
+const revertAgentVersion = async (searchParameter, versionIndex) => {
+  const agent = await Agent.findOne(searchParameter);
+  if (!agent) {
+    throw new Error('Agent not found');
+  }
+
+  if (!agent.versions || !agent.versions[versionIndex]) {
+    throw new Error(`Version ${versionIndex} not found`);
+  }
+
+  const revertToVersion = agent.versions[versionIndex];
+
+  const updateData = {
+    ...revertToVersion,
+  };
+
+  delete updateData._id;
+  delete updateData.id;
+  delete updateData.versions;
+  delete updateData.author;
+  delete updateData.updatedBy;
+
+  return Agent.findOneAndUpdate(searchParameter, updateData, { new: true }).lean();
+};
+
+/**
+ * Generates a hash of action metadata for version comparison
+ * @param {string[]} actionIds - Array of action IDs in format "domain_action_id"
+ * @param {Action[]} actions - Array of action documents
+ * @returns {Promise<string>} - SHA256 hash of the action metadata
+ */
+const generateActionMetadataHash = async (actionIds, actions) => {
+  if (!actionIds || actionIds.length === 0) {
+    return '';
+  }
+
+  // Create a map of action_id to metadata for quick lookup
+  const actionMap = new Map();
+  actions.forEach((action) => {
+    actionMap.set(action.action_id, action.metadata);
+  });
+
+  // Sort action IDs for consistent hashing
+  const sortedActionIds = [...actionIds].sort();
+
+  // Build a deterministic string representation of all action metadata
+  const metadataString = sortedActionIds
+    .map((actionFullId) => {
+      // Extract just the action_id part (after the delimiter)
+      const parts = actionFullId.split(actionDelimiter);
+      const actionId = parts[1];
+
+      const metadata = actionMap.get(actionId);
+      if (!metadata) {
+        return `${actionId}:null`;
+      }
+
+      // Sort metadata keys for deterministic output
+      const sortedKeys = Object.keys(metadata).sort();
+      const metadataStr = sortedKeys
+        .map((key) => `${key}:${JSON.stringify(metadata[key])}`)
+        .join(',');
+      return `${actionId}:{${metadataStr}}`;
+    })
+    .join(';');
+
+  // Use Web Crypto API to generate hash
+  const encoder = new TextEncoder();
+  const data = encoder.encode(metadataString);
+  const hashBuffer = await crypto.webcrypto.subtle.digest('SHA-256', data);
+  const hashArray = Array.from(new Uint8Array(hashBuffer));
+  const hashHex = hashArray.map((b) => b.toString(16).padStart(2, '0')).join('');
+
+  return hashHex;
+};
+
+/**
+ * Load a default agent based on the endpoint
+ * @param {string} endpoint
+ * @returns {Agent | null}
+ */
+
 module.exports = {
-  Agent,
  getAgent,
  loadAgent,
  createAgent,
  updateAgent,
  deleteAgent,
  getListAgents,
+  revertAgentVersion,
  updateAgentProjects,
  addAgentResourceFile,
  removeAgentResourceFiles,
+  generateActionMetadataHash,
 };
--- a/api/models/Agent.spec.js
+++ b/api/models/Agent.spec.js
--- a/api/models/Assistant.js
+++ b/api/models/Assistant.js
@@ -1,7 +1,4 @@
-const mongoose = require('mongoose');
-const assistantSchema = require('./schema/assistant');
-
-const Assistant = mongoose.model('assistant', assistantSchema);
+const { Assistant } = require('~/db/models');

 /**
 * Update an assistant with new data without overwriting existing properties,
--- a/api/models/Balance.js
+++ b/api/models/Balance.js
@@ -1,44 +0,0 @@
-const mongoose = require('mongoose');
-const balanceSchema = require('./schema/balance');
-const { getMultiplier } = require('./tx');
-const { logger } = require('~/config');
-
-balanceSchema.statics.check = async function ({
-  user,
-  model,
-  endpoint,
-  valueKey,
-  tokenType,
-  amount,
-  endpointTokenConfig,
-}) {
-  const multiplier = getMultiplier({ valueKey, tokenType, model, endpoint, endpointTokenConfig });
-  const tokenCost = amount * multiplier;
-  const { tokenCredits: balance } = (await this.findOne({ user }, 'tokenCredits').lean()) ?? {};
-
-  logger.debug('[Balance.check]', {
-    user,
-    model,
-    endpoint,
-    valueKey,
-    tokenType,
-    amount,
-    balance,
-    multiplier,
-    endpointTokenConfig: !!endpointTokenConfig,
-  });
-
-  if (!balance) {
-    return {
-      canSpend: false,
-      balance: 0,
-      tokenCost,
-    };
-  }
-
-  logger.debug('[Balance.check]', { tokenCost });
-
-  return { canSpend: balance >= tokenCost, balance, tokenCost };
-};
-
-module.exports = mongoose.model('Balance', balanceSchema);
--- a/api/models/Banner.js
+++ b/api/models/Banner.js
@@ -1,5 +1,6 @@
-const Banner = require('./schema/banner');
-const logger = require('~/config/winston');
+const { logger } = require('@librechat/data-schemas');
+const { Banner } = require('~/db/models');
+
 /**
 * Retrieves the current active banner.
 * @returns {Promise<Object|null>} The active banner object or null if no active banner is found.
--- a/api/models/Categories.js
+++ b/api/models/Categories.js
@@ -1,5 +1,4 @@
 const { logger } = require('~/config');
-// const { Categories } = require('./schema/categories');

 const options = [
  {
--- a/api/models/Config.js
+++ b/api/models/Config.js
@@ -1,86 +0,0 @@
-const mongoose = require('mongoose');
-const { logger } = require('~/config');
-
-const major = [0, 0];
-const minor = [0, 0];
-const patch = [0, 5];
-
-const configSchema = mongoose.Schema(
-  {
-    tag: {
-      type: String,
-      required: true,
-      validate: {
-        validator: function (tag) {
-          const [part1, part2, part3] = tag.replace('v', '').split('.').map(Number);
-
-          // Check if all parts are numbers
-          if (isNaN(part1) || isNaN(part2) || isNaN(part3)) {
-            return false;
-          }
-
-          // Check if all parts are within their respective ranges
-          if (part1 < major[0] || part1 > major[1]) {
-            return false;
-          }
-          if (part2 < minor[0] || part2 > minor[1]) {
-            return false;
-          }
-          if (part3 < patch[0] || part3 > patch[1]) {
-            return false;
-          }
-          return true;
-        },
-        message: 'Invalid tag value',
-      },
-    },
-    searchEnabled: {
-      type: Boolean,
-      default: false,
-    },
-    usersEnabled: {
-      type: Boolean,
-      default: false,
-    },
-    startupCounts: {
-      type: Number,
-      default: 0,
-    },
-  },
-  { timestamps: true },
-);
-
-// Instance method
-configSchema.methods.incrementCount = function () {
-  this.startupCounts += 1;
-};
-
-// Static methods
-configSchema.statics.findByTag = async function (tag) {
-  return await this.findOne({ tag }).lean();
-};
-
-configSchema.statics.updateByTag = async function (tag, update) {
-  return await this.findOneAndUpdate({ tag }, update, { new: true });
-};
-
-const Config = mongoose.models.Config || mongoose.model('Config', configSchema);
-
-module.exports = {
-  getConfigs: async (filter) => {
-    try {
-      return await Config.find(filter).lean();
-    } catch (error) {
-      logger.error('Error getting configs', error);
-      return { config: 'Error getting configs' };
-    }
-  },
-  deleteConfigs: async (filter) => {
-    try {
-      return await Config.deleteMany(filter);
-    } catch (error) {
-      logger.error('Error deleting configs', error);
-      return { config: 'Error deleting configs' };
-    }
-  },
-};
--- a/api/models/Conversation.js
+++ b/api/models/Conversation.js
@@ -1,6 +1,8 @@
-const Conversation = require('./schema/convoSchema');
+const { logger } = require('@librechat/data-schemas');
+const { createTempChatExpirationDate } = require('@librechat/api');
+const getCustomConfig = require('~/server/services/Config/getCustomConfig');
 const { getMessages, deleteMessages } = require('./Message');
-const logger = require('~/config/winston');
+const { Conversation } = require('~/db/models');

 /**
 * Searches for a conversation by conversationId and returns a lean document with only conversationId and user.
@@ -15,19 +17,6 @@ const searchConversation = async (conversationId) => {
    throw new Error('Error searching conversation');
  }
 };
-/**
- * Searches for a conversation by conversationId and returns associated file ids.
- * @param {string} conversationId - The conversation's ID.
- * @returns {Promise<string[] | null>}
- */
-const getConvoFiles = async (conversationId) => {
-  try {
-    return (await Conversation.findOne({ conversationId }, 'files').lean())?.files ?? [];
-  } catch (error) {
-    logger.error('[getConvoFiles] Error getting conversation files', error);
-    throw new Error('Error getting conversation files');
-  }
-};

 /**
 * Retrieves a single conversation for a given user and conversation ID.
@@ -73,8 +62,21 @@ const deleteNullOrEmptyConversations = async () => {
  }
 };

+/**
+ * Searches for a conversation by conversationId and returns associated file ids.
+ * @param {string} conversationId - The conversation's ID.
+ * @returns {Promise<string[] | null>}
+ */
+const getConvoFiles = async (conversationId) => {
+  try {
+    return (await Conversation.findOne({ conversationId }, 'files').lean())?.files ?? [];
+  } catch (error) {
+    logger.error('[getConvoFiles] Error getting conversation files', error);
+    throw new Error('Error getting conversation files');
+  }
+};
+
 module.exports = {
-  Conversation,
  getConvoFiles,
  searchConversation,
  deleteNullOrEmptyConversations,
@@ -87,27 +89,40 @@ module.exports = {
   */
  saveConvo: async (req, { conversationId, newConversationId, ...convo }, metadata) => {
    try {
-      if (metadata && metadata?.context) {
+      if (metadata?.context) {
        logger.debug(`[saveConvo] ${metadata.context}`);
      }
+
      const messages = await getMessages({ conversationId }, '_id');
      const update = { ...convo, messages, user: req.user.id };
+
      if (newConversationId) {
        update.conversationId = newConversationId;
      }

-      if (req.body.isTemporary) {
-        const expiredAt = new Date();
-        expiredAt.setDate(expiredAt.getDate() + 30);
-        update.expiredAt = expiredAt;
+      if (req?.body?.isTemporary) {
+        try {
+          const customConfig = await getCustomConfig();
+          update.expiredAt = createTempChatExpirationDate(customConfig);
+        } catch (err) {
+          logger.error('Error creating temporary chat expiration date:', err);
+          logger.info(`---\`saveConvo\` context: ${metadata?.context}`);
+          update.expiredAt = null;
+        }
      } else {
        update.expiredAt = null;
      }

+      /** @type {{ $set: Partial<TConversation>; $unset?: Record<keyof TConversation, number> }} */
+      const updateOperation = { $set: update };
+      if (metadata && metadata.unsetFields && Object.keys(metadata.unsetFields).length > 0) {
+        updateOperation.$unset = metadata.unsetFields;
+      }
+
      /** Note: the resulting Model object is necessary for Meilisearch operations */
      const conversation = await Conversation.findOneAndUpdate(
        { conversationId, user: req.user.id },
-        update,
+        updateOperation,
        {
          new: true,
          upsert: true,
@@ -141,75 +156,101 @@ module.exports = {
      throw new Error('Failed to save conversations in bulk.');
    }
  },
-  getConvosByPage: async (user, pageNumber = 1, pageSize = 25, isArchived = false, tags) => {
-    const query = { user };
+  getConvosByCursor: async (
+    user,
+    { cursor, limit = 25, isArchived = false, tags, search, order = 'desc' } = {},
+  ) => {
+    const filters = [{ user }];
    if (isArchived) {
-      query.isArchived = true;
+      filters.push({ isArchived: true });
    } else {
-      query.$or = [{ isArchived: false }, { isArchived: { $exists: false } }];
-    }
-    if (Array.isArray(tags) && tags.length > 0) {
-      query.tags = { $in: tags };
+      filters.push({ $or: [{ isArchived: false }, { isArchived: { $exists: false } }] });
    }

-    query.$and = [{ $or: [{ expiredAt: null }, { expiredAt: { $exists: false } }] }];
+    if (Array.isArray(tags) && tags.length > 0) {
+      filters.push({ tags: { $in: tags } });
+    }
+
+    filters.push({ $or: [{ expiredAt: null }, { expiredAt: { $exists: false } }] });
+
+    if (search) {
+      try {
+        const meiliResults = await Conversation.meiliSearch(search);
+        const matchingIds = Array.isArray(meiliResults.hits)
+          ? meiliResults.hits.map((result) => result.conversationId)
+          : [];
+        if (!matchingIds.length) {
+          return { conversations: [], nextCursor: null };
+        }
+        filters.push({ conversationId: { $in: matchingIds } });
+      } catch (error) {
+        logger.error('[getConvosByCursor] Error during meiliSearch', error);
+        return { message: 'Error during meiliSearch' };
+      }
+    }
+
+    if (cursor) {
+      filters.push({ updatedAt: { $lt: new Date(cursor) } });
+    }
+
+    const query = filters.length === 1 ? filters[0] : { $and: filters };

    try {
-      const totalConvos = (await Conversation.countDocuments(query)) || 1;
-      const totalPages = Math.ceil(totalConvos / pageSize);
      const convos = await Conversation.find(query)
-        .sort({ updatedAt: -1 })
-        .skip((pageNumber - 1) * pageSize)
-        .limit(pageSize)
+        .select(
+          'conversationId endpoint title createdAt updatedAt user model agent_id assistant_id spec iconURL',
+        )
+        .sort({ updatedAt: order === 'asc' ? 1 : -1 })
+        .limit(limit + 1)
        .lean();
-      return { conversations: convos, pages: totalPages, pageNumber, pageSize };
+
+      let nextCursor = null;
+      if (convos.length > limit) {
+        const lastConvo = convos.pop();
+        nextCursor = lastConvo.updatedAt.toISOString();
+      }
+
+      return { conversations: convos, nextCursor };
    } catch (error) {
-      logger.error('[getConvosByPage] Error getting conversations', error);
+      logger.error('[getConvosByCursor] Error getting conversations', error);
      return { message: 'Error getting conversations' };
    }
  },
-  getConvosQueried: async (user, convoIds, pageNumber = 1, pageSize = 25) => {
+  getConvosQueried: async (user, convoIds, cursor = null, limit = 25) => {
    try {
-      if (!convoIds || convoIds.length === 0) {
-        return { conversations: [], pages: 1, pageNumber, pageSize };
+      if (!convoIds?.length) {
+        return { conversations: [], nextCursor: null, convoMap: {} };
+      }
+
+      const conversationIds = convoIds.map((convo) => convo.conversationId);
+
+      const results = await Conversation.find({
+        user,
+        conversationId: { $in: conversationIds },
+        $or: [{ expiredAt: { $exists: false } }, { expiredAt: null }],
+      }).lean();
+
+      results.sort((a, b) => new Date(b.updatedAt) - new Date(a.updatedAt));
+
+      let filtered = results;
+      if (cursor && cursor !== 'start') {
+        const cursorDate = new Date(cursor);
+        filtered = results.filter((convo) => new Date(convo.updatedAt) < cursorDate);
+      }
+
+      const limited = filtered.slice(0, limit + 1);
+      let nextCursor = null;
+      if (limited.length > limit) {
+        const lastConvo = limited.pop();
+        nextCursor = lastConvo.updatedAt.toISOString();
      }

-      const cache = {};
      const convoMap = {};
-      const promises = [];
-
-      convoIds.forEach((convo) =>
-        promises.push(
-          Conversation.findOne({
-            user,
-            conversationId: convo.conversationId,
-            $or: [{ expiredAt: { $exists: false } }, { expiredAt: null }],
-          }).lean(),
-        ),
-      );
-
-      const results = (await Promise.all(promises)).filter(Boolean);
-
-      results.forEach((convo, i) => {
-        const page = Math.floor(i / pageSize) + 1;
-        if (!cache[page]) {
-          cache[page] = [];
-        }
-        cache[page].push(convo);
+      limited.forEach((convo) => {
        convoMap[convo.conversationId] = convo;
      });

-      const totalPages = Math.ceil(results.length / pageSize);
-      cache.pages = totalPages;
-      cache.pageSize = pageSize;
-      return {
-        cache,
-        conversations: cache[pageNumber] || [],
-        pages: totalPages || 1,
-        pageNumber,
-        pageSize,
-        convoMap,
-      };
+      return { conversations: limited, nextCursor, convoMap };
    } catch (error) {
      logger.error('[getConvosQueried] Error getting conversations', error);
      return { message: 'Error fetching conversations' };
@@ -250,10 +291,25 @@ module.exports = {
   * logger.error(result); // { n: 5, ok: 1, deletedCount: 5, messages: { n: 10, ok: 1, deletedCount: 10 } }
   */
  deleteConvos: async (user, filter) => {
-    let toRemove = await Conversation.find({ ...filter, user }).select('conversationId');
-    const ids = toRemove.map((instance) => instance.conversationId);
-    let deleteCount = await Conversation.deleteMany({ ...filter, user });
-    deleteCount.messages = await deleteMessages({ conversationId: { $in: ids } });
-    return deleteCount;
+    try {
+      const userFilter = { ...filter, user };
+      const conversations = await Conversation.find(userFilter).select('conversationId');
+      const conversationIds = conversations.map((c) => c.conversationId);
+
+      if (!conversationIds.length) {
+        throw new Error('Conversation not found or already deleted.');
+      }
+
+      const deleteConvoResult = await Conversation.deleteMany(userFilter);
+
+      const deleteMessagesResult = await deleteMessages({
+        conversationId: { $in: conversationIds },
+      });
+
+      return { ...deleteConvoResult, messages: deleteMessagesResult };
+    } catch (error) {
+      logger.error('[deleteConvos] Error deleting conversations and messages', error);
+      throw error;
+    }
  },
 };
--- a/api/models/ConversationTag.js
+++ b/api/models/ConversationTag.js
@@ -1,6 +1,5 @@
-const ConversationTag = require('./schema/conversationTagSchema');
-const Conversation = require('./schema/convoSchema');
-const logger = require('~/config/winston');
+const { logger } = require('@librechat/data-schemas');
+const { ConversationTag, Conversation } = require('~/db/models');

 /**
 * Retrieves all conversation tags for a user.
@@ -136,13 +135,13 @@ const adjustPositions = async (user, oldPosition, newPosition) => {
  const position =
    oldPosition < newPosition
      ? {
-        $gt: Math.min(oldPosition, newPosition),
-        $lte: Math.max(oldPosition, newPosition),
-      }
+          $gt: Math.min(oldPosition, newPosition),
+          $lte: Math.max(oldPosition, newPosition),
+        }
      : {
-        $gte: Math.min(oldPosition, newPosition),
-        $lt: Math.max(oldPosition, newPosition),
-      };
+          $gte: Math.min(oldPosition, newPosition),
+          $lt: Math.max(oldPosition, newPosition),
+        };

  await ConversationTag.updateMany(
    {
--- a/api/models/File.js
+++ b/api/models/File.js
@@ -1,7 +1,8 @@
-const mongoose = require('mongoose');
-const fileSchema = require('./schema/fileSchema');
-
-const File = mongoose.model('File', fileSchema);
+const { logger } = require('@librechat/data-schemas');
+const { EToolResources, FileContext, Constants } = require('librechat-data-provider');
+const { getProjectByName } = require('./Project');
+const { getAgent } = require('./Agent');
+const { File } = require('~/db/models');

 /**
 * Finds a file by its file_id with additional query options.
@@ -13,15 +14,161 @@ const findFileById = async (file_id, options = {}) => {
  return await File.findOne({ file_id, ...options }).lean();
 };

+/**
+ * Checks if a user has access to multiple files through a shared agent (batch operation)
+ * @param {string} userId - The user ID to check access for
+ * @param {string[]} fileIds - Array of file IDs to check
+ * @param {string} agentId - The agent ID that might grant access
+ * @returns {Promise<Map<string, boolean>>} Map of fileId to access status
+ */
+const hasAccessToFilesViaAgent = async (userId, fileIds, agentId, checkCollaborative = true) => {
+  const accessMap = new Map();
+
+  // Initialize all files as no access
+  fileIds.forEach((fileId) => accessMap.set(fileId, false));
+
+  try {
+    const agent = await getAgent({ id: agentId });
+
+    if (!agent) {
+      return accessMap;
+    }
+
+    // Check if user is the author - if so, grant access to all files
+    if (agent.author.toString() === userId) {
+      fileIds.forEach((fileId) => accessMap.set(fileId, true));
+      return accessMap;
+    }
+
+    // Check if agent is shared with the user via projects
+    if (!agent.projectIds || agent.projectIds.length === 0) {
+      return accessMap;
+    }
+
+    // Check if agent is in global project
+    const globalProject = await getProjectByName(Constants.GLOBAL_PROJECT_NAME, '_id');
+    if (
+      !globalProject ||
+      !agent.projectIds.some((pid) => pid.toString() === globalProject._id.toString())
+    ) {
+      return accessMap;
+    }
+
+    // Agent is globally shared - check if it's collaborative
+    if (checkCollaborative && !agent.isCollaborative) {
+      return accessMap;
+    }
+
+    // Check which files are actually attached
+    const attachedFileIds = new Set();
+    if (agent.tool_resources) {
+      for (const [_resourceType, resource] of Object.entries(agent.tool_resources)) {
+        if (resource?.file_ids && Array.isArray(resource.file_ids)) {
+          resource.file_ids.forEach((fileId) => attachedFileIds.add(fileId));
+        }
+      }
+    }
+
+    // Grant access only to files that are attached to this agent
+    fileIds.forEach((fileId) => {
+      if (attachedFileIds.has(fileId)) {
+        accessMap.set(fileId, true);
+      }
+    });
+
+    return accessMap;
+  } catch (error) {
+    logger.error('[hasAccessToFilesViaAgent] Error checking file access:', error);
+    return accessMap;
+  }
+};
+
 /**
 * Retrieves files matching a given filter, sorted by the most recently updated.
 * @param {Object} filter - The filter criteria to apply.
 * @param {Object} [_sortOptions] - Optional sort parameters.
+ * @param {Object|String} [selectFields={ text: 0 }] - Fields to include/exclude in the query results.
+ *                                                   Default excludes the 'text' field.
+ * @param {Object} [options] - Additional options
+ * @param {string} [options.userId] - User ID for access control
+ * @param {string} [options.agentId] - Agent ID that might grant access to files
 * @returns {Promise<Array<MongoFile>>} A promise that resolves to an array of file documents.
 */
-const getFiles = async (filter, _sortOptions) => {
+const getFiles = async (filter, _sortOptions, selectFields = { text: 0 }, options = {}) => {
  const sortOptions = { updatedAt: -1, ..._sortOptions };
-  return await File.find(filter).sort(sortOptions).lean();
+  const files = await File.find(filter).select(selectFields).sort(sortOptions).lean();
+
+  // If userId and agentId are provided, filter files based on access
+  if (options.userId && options.agentId) {
+    // Collect file IDs that need access check
+    const filesToCheck = [];
+    const ownedFiles = [];
+
+    for (const file of files) {
+      if (file.user && file.user.toString() === options.userId) {
+        ownedFiles.push(file);
+      } else {
+        filesToCheck.push(file);
+      }
+    }
+
+    if (filesToCheck.length === 0) {
+      return ownedFiles;
+    }
+
+    // Batch check access for all non-owned files
+    const fileIds = filesToCheck.map((f) => f.file_id);
+    const accessMap = await hasAccessToFilesViaAgent(
+      options.userId,
+      fileIds,
+      options.agentId,
+      false,
+    );
+
+    // Filter files based on access
+    const accessibleFiles = filesToCheck.filter((file) => accessMap.get(file.file_id));
+
+    return [...ownedFiles, ...accessibleFiles];
+  }
+
+  return files;
+};
+
+/**
+ * Retrieves tool files (files that are embedded or have a fileIdentifier) from an array of file IDs
+ * @param {string[]} fileIds - Array of file_id strings to search for
+ * @param {Set<EToolResources>} toolResourceSet - Optional filter for tool resources
+ * @returns {Promise<Array<MongoFile>>} Files that match the criteria
+ */
+const getToolFilesByIds = async (fileIds, toolResourceSet) => {
+  if (!fileIds || !fileIds.length || !toolResourceSet?.size) {
+    return [];
+  }
+
+  try {
+    const filter = {
+      file_id: { $in: fileIds },
+      $or: [],
+    };
+
+    if (toolResourceSet.has(EToolResources.ocr)) {
+      filter.$or.push({ text: { $exists: true, $ne: null }, context: FileContext.agents });
+    }
+    if (toolResourceSet.has(EToolResources.file_search)) {
+      filter.$or.push({ embedded: true });
+    }
+    if (toolResourceSet.has(EToolResources.execute_code)) {
+      filter.$or.push({ 'metadata.fileIdentifier': { $exists: true } });
+    }
+
+    const selectFields = { text: 0 };
+    const sortOptions = { updatedAt: -1 };
+
+    return await getFiles(filter, sortOptions, selectFields);
+  } catch (error) {
+    logger.error('[getToolFilesByIds] Error retrieving tool files:', error);
+    throw new Error('Error retrieving tool files');
+  }
 };

 /**
@@ -105,14 +252,38 @@ const deleteFiles = async (file_ids, user) => {
  return await File.deleteMany(deleteQuery);
 };

+/**
+ * Batch updates files with new signed URLs in MongoDB
+ *
+ * @param {MongoFile[]} updates - Array of updates in the format { file_id, filepath }
+ * @returns {Promise<void>}
+ */
+async function batchUpdateFiles(updates) {
+  if (!updates || updates.length === 0) {
+    return;
+  }
+
+  const bulkOperations = updates.map((update) => ({
+    updateOne: {
+      filter: { file_id: update.file_id },
+      update: { $set: { filepath: update.filepath } },
+    },
+  }));
+
+  const result = await File.bulkWrite(bulkOperations);
+  logger.info(`Updated ${result.modifiedCount} files with new S3 URLs`);
+}
+
 module.exports = {
-  File,
  findFileById,
  getFiles,
+  getToolFilesByIds,
  createFile,
  updateFile,
  updateFileUsage,
  deleteFile,
  deleteFiles,
  deleteFileByFilter,
+  batchUpdateFiles,
+  hasAccessToFilesViaAgent,
 };
--- a/api/models/File.spec.js
+++ b/api/models/File.spec.js
@@ -0,0 +1,264 @@
+const mongoose = require('mongoose');
+const { v4: uuidv4 } = require('uuid');
+const { fileSchema } = require('@librechat/data-schemas');
+const { agentSchema } = require('@librechat/data-schemas');
+const { projectSchema } = require('@librechat/data-schemas');
+const { MongoMemoryServer } = require('mongodb-memory-server');
+const { GLOBAL_PROJECT_NAME } = require('librechat-data-provider').Constants;
+const { getFiles, createFile } = require('./File');
+const { getProjectByName } = require('./Project');
+const { createAgent } = require('./Agent');
+
+let File;
+let Agent;
+let Project;
+
+describe('File Access Control', () => {
+  let mongoServer;
+
+  beforeAll(async () => {
+    mongoServer = await MongoMemoryServer.create();
+    const mongoUri = mongoServer.getUri();
+    File = mongoose.models.File || mongoose.model('File', fileSchema);
+    Agent = mongoose.models.Agent || mongoose.model('Agent', agentSchema);
+    Project = mongoose.models.Project || mongoose.model('Project', projectSchema);
+    await mongoose.connect(mongoUri);
+  });
+
+  afterAll(async () => {
+    await mongoose.disconnect();
+    await mongoServer.stop();
+  });
+
+  beforeEach(async () => {
+    await File.deleteMany({});
+    await Agent.deleteMany({});
+    await Project.deleteMany({});
+  });
+
+  describe('hasAccessToFilesViaAgent', () => {
+    it('should efficiently check access for multiple files at once', async () => {
+      const userId = new mongoose.Types.ObjectId().toString();
+      const authorId = new mongoose.Types.ObjectId().toString();
+      const agentId = uuidv4();
+      const fileIds = [uuidv4(), uuidv4(), uuidv4(), uuidv4()];
+
+      // Create files
+      for (const fileId of fileIds) {
+        await createFile({
+          user: authorId,
+          file_id: fileId,
+          filename: `file-${fileId}.txt`,
+          filepath: `/uploads/${fileId}`,
+        });
+      }
+
+      // Create agent with only first two files attached
+      await createAgent({
+        id: agentId,
+        name: 'Test Agent',
+        author: authorId,
+        model: 'gpt-4',
+        provider: 'openai',
+        isCollaborative: true,
+        tool_resources: {
+          file_search: {
+            file_ids: [fileIds[0], fileIds[1]],
+          },
+        },
+      });
+
+      // Get or create global project
+      const globalProject = await getProjectByName(GLOBAL_PROJECT_NAME, '_id');
+
+      // Share agent globally
+      await Agent.updateOne({ id: agentId }, { $push: { projectIds: globalProject._id } });
+
+      // Check access for all files
+      const { hasAccessToFilesViaAgent } = require('./File');
+      const accessMap = await hasAccessToFilesViaAgent(userId, fileIds, agentId);
+
+      // Should have access only to the first two files
+      expect(accessMap.get(fileIds[0])).toBe(true);
+      expect(accessMap.get(fileIds[1])).toBe(true);
+      expect(accessMap.get(fileIds[2])).toBe(false);
+      expect(accessMap.get(fileIds[3])).toBe(false);
+    });
+
+    it('should grant access to all files when user is the agent author', async () => {
+      const authorId = new mongoose.Types.ObjectId().toString();
+      const agentId = uuidv4();
+      const fileIds = [uuidv4(), uuidv4(), uuidv4()];
+
+      // Create agent
+      await createAgent({
+        id: agentId,
+        name: 'Test Agent',
+        author: authorId,
+        model: 'gpt-4',
+        provider: 'openai',
+        tool_resources: {
+          file_search: {
+            file_ids: [fileIds[0]], // Only one file attached
+          },
+        },
+      });
+
+      // Check access as the author
+      const { hasAccessToFilesViaAgent } = require('./File');
+      const accessMap = await hasAccessToFilesViaAgent(authorId, fileIds, agentId);
+
+      // Author should have access to all files
+      expect(accessMap.get(fileIds[0])).toBe(true);
+      expect(accessMap.get(fileIds[1])).toBe(true);
+      expect(accessMap.get(fileIds[2])).toBe(true);
+    });
+
+    it('should handle non-existent agent gracefully', async () => {
+      const userId = new mongoose.Types.ObjectId().toString();
+      const fileIds = [uuidv4(), uuidv4()];
+
+      const { hasAccessToFilesViaAgent } = require('./File');
+      const accessMap = await hasAccessToFilesViaAgent(userId, fileIds, 'non-existent-agent');
+
+      // Should have no access to any files
+      expect(accessMap.get(fileIds[0])).toBe(false);
+      expect(accessMap.get(fileIds[1])).toBe(false);
+    });
+
+    it('should deny access when agent is not collaborative', async () => {
+      const userId = new mongoose.Types.ObjectId().toString();
+      const authorId = new mongoose.Types.ObjectId().toString();
+      const agentId = uuidv4();
+      const fileIds = [uuidv4(), uuidv4()];
+
+      // Create agent with files but isCollaborative: false
+      await createAgent({
+        id: agentId,
+        name: 'Non-Collaborative Agent',
+        author: authorId,
+        model: 'gpt-4',
+        provider: 'openai',
+        isCollaborative: false,
+        tool_resources: {
+          file_search: {
+            file_ids: fileIds,
+          },
+        },
+      });
+
+      // Get or create global project
+      const globalProject = await getProjectByName(GLOBAL_PROJECT_NAME, '_id');
+
+      // Share agent globally
+      await Agent.updateOne({ id: agentId }, { $push: { projectIds: globalProject._id } });
+
+      // Check access for files
+      const { hasAccessToFilesViaAgent } = require('./File');
+      const accessMap = await hasAccessToFilesViaAgent(userId, fileIds, agentId);
+
+      // Should have no access to any files when isCollaborative is false
+      expect(accessMap.get(fileIds[0])).toBe(false);
+      expect(accessMap.get(fileIds[1])).toBe(false);
+    });
+  });
+
+  describe('getFiles with agent access control', () => {
+    test('should return files owned by user and files accessible through agent', async () => {
+      const authorId = new mongoose.Types.ObjectId();
+      const userId = new mongoose.Types.ObjectId();
+      const agentId = `agent_${uuidv4()}`;
+      const ownedFileId = `file_${uuidv4()}`;
+      const sharedFileId = `file_${uuidv4()}`;
+      const inaccessibleFileId = `file_${uuidv4()}`;
+
+      // Create/get global project using getProjectByName which will upsert
+      const globalProject = await getProjectByName(GLOBAL_PROJECT_NAME);
+
+      // Create agent with shared file
+      await createAgent({
+        id: agentId,
+        name: 'Shared Agent',
+        provider: 'test',
+        model: 'test-model',
+        author: authorId,
+        projectIds: [globalProject._id],
+        isCollaborative: true,
+        tool_resources: {
+          file_search: {
+            file_ids: [sharedFileId],
+          },
+        },
+      });
+
+      // Create files
+      await createFile({
+        file_id: ownedFileId,
+        user: userId,
+        filename: 'owned.txt',
+        filepath: '/uploads/owned.txt',
+        type: 'text/plain',
+        bytes: 100,
+      });
+
+      await createFile({
+        file_id: sharedFileId,
+        user: authorId,
+        filename: 'shared.txt',
+        filepath: '/uploads/shared.txt',
+        type: 'text/plain',
+        bytes: 200,
+        embedded: true,
+      });
+
+      await createFile({
+        file_id: inaccessibleFileId,
+        user: authorId,
+        filename: 'inaccessible.txt',
+        filepath: '/uploads/inaccessible.txt',
+        type: 'text/plain',
+        bytes: 300,
+      });
+
+      // Get files with access control
+      const files = await getFiles(
+        { file_id: { $in: [ownedFileId, sharedFileId, inaccessibleFileId] } },
+        null,
+        { text: 0 },
+        { userId: userId.toString(), agentId },
+      );
+
+      expect(files).toHaveLength(2);
+      expect(files.map((f) => f.file_id)).toContain(ownedFileId);
+      expect(files.map((f) => f.file_id)).toContain(sharedFileId);
+      expect(files.map((f) => f.file_id)).not.toContain(inaccessibleFileId);
+    });
+
+    test('should return all files when no userId/agentId provided', async () => {
+      const userId = new mongoose.Types.ObjectId();
+      const fileId1 = `file_${uuidv4()}`;
+      const fileId2 = `file_${uuidv4()}`;
+
+      await createFile({
+        file_id: fileId1,
+        user: userId,
+        filename: 'file1.txt',
+        filepath: '/uploads/file1.txt',
+        type: 'text/plain',
+        bytes: 100,
+      });
+
+      await createFile({
+        file_id: fileId2,
+        user: new mongoose.Types.ObjectId(),
+        filename: 'file2.txt',
+        filepath: '/uploads/file2.txt',
+        type: 'text/plain',
+        bytes: 200,
+      });
+
+      const files = await getFiles({ file_id: { $in: [fileId1, fileId2] } });
+      expect(files).toHaveLength(2);
+    });
+  });
+});
--- a/api/models/Key.js
+++ b/api/models/Key.js
@@ -1,4 +0,0 @@
-const mongoose = require('mongoose');
-const keySchema = require('./schema/key');
-
-module.exports = mongoose.model('Key', keySchema);
--- a/api/models/Message.js
+++ b/api/models/Message.js
@@ -1,6 +1,8 @@
 const { z } = require('zod');
-const Message = require('./schema/messageSchema');
-const { logger } = require('~/config');
+const { logger } = require('@librechat/data-schemas');
+const { createTempChatExpirationDate } = require('@librechat/api');
+const getCustomConfig = require('~/server/services/Config/getCustomConfig');
+const { Message } = require('~/db/models');

 const idSchema = z.string().uuid();

@@ -54,13 +56,25 @@ async function saveMessage(req, params, metadata) {
    };

    if (req?.body?.isTemporary) {
-      const expiredAt = new Date();
-      expiredAt.setDate(expiredAt.getDate() + 30);
-      update.expiredAt = expiredAt;
+      try {
+        const customConfig = await getCustomConfig();
+        update.expiredAt = createTempChatExpirationDate(customConfig);
+      } catch (err) {
+        logger.error('Error creating temporary chat expiration date:', err);
+        logger.info(`---\`saveMessage\` context: ${metadata?.context}`);
+        update.expiredAt = null;
+      }
    } else {
      update.expiredAt = null;
    }

+    if (update.tokenCount != null && isNaN(update.tokenCount)) {
+      logger.warn(
+        `Resetting invalid \`tokenCount\` for message \`${params.messageId}\`: ${update.tokenCount}`,
+      );
+      logger.info(`---\`saveMessage\` context: ${metadata?.context}`);
+      update.tokenCount = 0;
+    }
    const message = await Message.findOneAndUpdate(
      { messageId: params.messageId, user: req.user.id },
      update,
@@ -71,7 +85,44 @@ async function saveMessage(req, params, metadata) {
  } catch (err) {
    logger.error('Error saving message:', err);
    logger.info(`---\`saveMessage\` context: ${metadata?.context}`);
-    throw err;
+
+    // Check if this is a duplicate key error (MongoDB error code 11000)
+    if (err.code === 11000 && err.message.includes('duplicate key error')) {
+      // Log the duplicate key error but don't crash the application
+      logger.warn(`Duplicate messageId detected: ${params.messageId}. Continuing execution.`);
+
+      try {
+        // Try to find the existing message with this ID
+        const existingMessage = await Message.findOne({
+          messageId: params.messageId,
+          user: req.user.id,
+        });
+
+        // If we found it, return it
+        if (existingMessage) {
+          return existingMessage.toObject();
+        }
+
+        // If we can't find it (unlikely but possible in race conditions)
+        return {
+          ...params,
+          messageId: params.messageId,
+          user: req.user.id,
+        };
+      } catch (findError) {
+        // If the findOne also fails, log it but don't crash
+        logger.warn(
+          `Could not retrieve existing message with ID ${params.messageId}: ${findError.message}`,
+        );
+        return {
+          ...params,
+          messageId: params.messageId,
+          user: req.user.id,
+        };
+      }
+    }
+
+    throw err; // Re-throw other errors
  }
 }

@@ -95,7 +146,6 @@ async function bulkSaveMessages(messages, overrideTimestamp = false) {
        upsert: true,
      },
    }));
-
    const result = await Message.bulkWrite(bulkOps);
    return result;
  } catch (err) {
@@ -210,6 +260,7 @@ async function updateMessage(req, message, metadata) {
      text: updatedMessage.text,
      isCreatedByUser: updatedMessage.isCreatedByUser,
      tokenCount: updatedMessage.tokenCount,
+      feedback: updatedMessage.feedback,
    };
  } catch (err) {
    logger.error('Error updating message:', err);
@@ -310,7 +361,6 @@ async function deleteMessages(filter) {
 }

 module.exports = {
-  Message,
  saveMessage,
  bulkSaveMessages,
  recordMessage,
--- a/api/models/Message.spec.js
+++ b/api/models/Message.spec.js
@@ -1,32 +1,7 @@
 const mongoose = require('mongoose');
+const { MongoMemoryServer } = require('mongodb-memory-server');
 const { v4: uuidv4 } = require('uuid');
-
-jest.mock('mongoose');
-
-const mockFindQuery = {
-  select: jest.fn().mockReturnThis(),
-  sort: jest.fn().mockReturnThis(),
-  lean: jest.fn().mockReturnThis(),
-  deleteMany: jest.fn().mockResolvedValue({ deletedCount: 1 }),
-};
-
-const mockSchema = {
-  findOneAndUpdate: jest.fn(),
-  updateOne: jest.fn(),
-  findOne: jest.fn(() => ({
-    lean: jest.fn(),
-  })),
-  find: jest.fn(() => mockFindQuery),
-  deleteMany: jest.fn(),
-};
-
-mongoose.model.mockReturnValue(mockSchema);
-
-jest.mock('~/models/schema/messageSchema', () => mockSchema);
-
-jest.mock('~/config/winston', () => ({
-  error: jest.fn(),
-}));
+const { messageSchema } = require('@librechat/data-schemas');

 const {
  saveMessage,
@@ -35,77 +10,102 @@ const {
  deleteMessages,
  updateMessageText,
  deleteMessagesSince,
-} = require('~/models/Message');
+} = require('./Message');
+
+/**
+ * @type {import('mongoose').Model<import('@librechat/data-schemas').IMessage>}
+ */
+let Message;

 describe('Message Operations', () => {
+  let mongoServer;
  let mockReq;
-  let mockMessage;
+  let mockMessageData;

-  beforeEach(() => {
-    jest.clearAllMocks();
+  beforeAll(async () => {
+    mongoServer = await MongoMemoryServer.create();
+    const mongoUri = mongoServer.getUri();
+    Message = mongoose.models.Message || mongoose.model('Message', messageSchema);
+    await mongoose.connect(mongoUri);
+  });
+
+  afterAll(async () => {
+    await mongoose.disconnect();
+    await mongoServer.stop();
+  });
+
+  beforeEach(async () => {
+    // Clear database
+    await Message.deleteMany({});

    mockReq = {
      user: { id: 'user123' },
    };

-    mockMessage = {
+    mockMessageData = {
      messageId: 'msg123',
      conversationId: uuidv4(),
      text: 'Hello, world!',
      user: 'user123',
    };
-
-    mockSchema.findOneAndUpdate.mockResolvedValue({
-      toObject: () => mockMessage,
-    });
  });

  describe('saveMessage', () => {
    it('should save a message for an authenticated user', async () => {
-      const result = await saveMessage(mockReq, mockMessage);
-      expect(result).toEqual(mockMessage);
-      expect(mockSchema.findOneAndUpdate).toHaveBeenCalledWith(
-        { messageId: 'msg123', user: 'user123' },
-        expect.objectContaining({ user: 'user123' }),
-        expect.any(Object),
-      );
+      const result = await saveMessage(mockReq, mockMessageData);
+
+      expect(result.messageId).toBe('msg123');
+      expect(result.user).toBe('user123');
+      expect(result.text).toBe('Hello, world!');
+
+      // Verify the message was actually saved to the database
+      const savedMessage = await Message.findOne({ messageId: 'msg123', user: 'user123' });
+      expect(savedMessage).toBeTruthy();
+      expect(savedMessage.text).toBe('Hello, world!');
    });

    it('should throw an error for unauthenticated user', async () => {
      mockReq.user = null;
-      await expect(saveMessage(mockReq, mockMessage)).rejects.toThrow('User not authenticated');
+      await expect(saveMessage(mockReq, mockMessageData)).rejects.toThrow('User not authenticated');
    });

-    it('should throw an error for invalid conversation ID', async () => {
-      mockMessage.conversationId = 'invalid-id';
-      await expect(saveMessage(mockReq, mockMessage)).resolves.toBeUndefined();
+    it('should handle invalid conversation ID gracefully', async () => {
+      mockMessageData.conversationId = 'invalid-id';
+      const result = await saveMessage(mockReq, mockMessageData);
+      expect(result).toBeUndefined();
    });
  });

  describe('updateMessageText', () => {
    it('should update message text for the authenticated user', async () => {
+      // First save a message
+      await saveMessage(mockReq, mockMessageData);
+
+      // Then update it
      await updateMessageText(mockReq, { messageId: 'msg123', text: 'Updated text' });
-      expect(mockSchema.updateOne).toHaveBeenCalledWith(
-        { messageId: 'msg123', user: 'user123' },
-        { text: 'Updated text' },
-      );
+
+      // Verify the update
+      const updatedMessage = await Message.findOne({ messageId: 'msg123', user: 'user123' });
+      expect(updatedMessage.text).toBe('Updated text');
    });
  });

  describe('updateMessage', () => {
    it('should update a message for the authenticated user', async () => {
-      mockSchema.findOneAndUpdate.mockResolvedValue(mockMessage);
+      // First save a message
+      await saveMessage(mockReq, mockMessageData);
+
      const result = await updateMessage(mockReq, { messageId: 'msg123', text: 'Updated text' });
-      expect(result).toEqual(
-        expect.objectContaining({
-          messageId: 'msg123',
-          text: 'Hello, world!',
-        }),
-      );
+
+      expect(result.messageId).toBe('msg123');
+      expect(result.text).toBe('Updated text');
+
+      // Verify in database
+      const updatedMessage = await Message.findOne({ messageId: 'msg123', user: 'user123' });
+      expect(updatedMessage.text).toBe('Updated text');
    });

    it('should throw an error if message is not found', async () => {
-      mockSchema.findOneAndUpdate.mockResolvedValue(null);
      await expect(
        updateMessage(mockReq, { messageId: 'nonexistent', text: 'Test' }),
      ).rejects.toThrow('Message not found or user not authorized.');
@@ -114,19 +114,45 @@ describe('Message Operations', () => {

  describe('deleteMessagesSince', () => {
    it('should delete messages only for the authenticated user', async () => {
-      mockSchema.findOne().lean.mockResolvedValueOnce({ createdAt: new Date() });
-      mockFindQuery.deleteMany.mockResolvedValueOnce({ deletedCount: 1 });
-      const result = await deleteMessagesSince(mockReq, {
-        messageId: 'msg123',
-        conversationId: 'convo123',
+      const conversationId = uuidv4();
+
+      // Create multiple messages in the same conversation
+      const message1 = await saveMessage(mockReq, {
+        messageId: 'msg1',
+        conversationId,
+        text: 'First message',
+        user: 'user123',
      });
-      expect(mockSchema.findOne).toHaveBeenCalledWith({ messageId: 'msg123', user: 'user123' });
-      expect(mockSchema.find).not.toHaveBeenCalled();
-      expect(result).toBeUndefined();
+
+      const message2 = await saveMessage(mockReq, {
+        messageId: 'msg2',
+        conversationId,
+        text: 'Second message',
+        user: 'user123',
+      });
+
+      const message3 = await saveMessage(mockReq, {
+        messageId: 'msg3',
+        conversationId,
+        text: 'Third message',
+        user: 'user123',
+      });
+
+      // Delete messages since message2 (this should only delete messages created AFTER msg2)
+      await deleteMessagesSince(mockReq, {
+        messageId: 'msg2',
+        conversationId,
+      });
+
+      // Verify msg1 and msg2 remain, msg3 is deleted
+      const remainingMessages = await Message.find({ conversationId, user: 'user123' });
+      expect(remainingMessages).toHaveLength(2);
+      expect(remainingMessages.map((m) => m.messageId)).toContain('msg1');
+      expect(remainingMessages.map((m) => m.messageId)).toContain('msg2');
+      expect(remainingMessages.map((m) => m.messageId)).not.toContain('msg3');
    });

    it('should return undefined if no message is found', async () => {
-      mockSchema.findOne().lean.mockResolvedValueOnce(null);
      const result = await deleteMessagesSince(mockReq, {
        messageId: 'nonexistent',
        conversationId: 'convo123',
@@ -137,29 +163,71 @@ describe('Message Operations', () => {

  describe('getMessages', () => {
    it('should retrieve messages with the correct filter', async () => {
-      const filter = { conversationId: 'convo123' };
-      await getMessages(filter);
-      expect(mockSchema.find).toHaveBeenCalledWith(filter);
-      expect(mockFindQuery.sort).toHaveBeenCalledWith({ createdAt: 1 });
-      expect(mockFindQuery.lean).toHaveBeenCalled();
+      const conversationId = uuidv4();
+
+      // Save some messages
+      await saveMessage(mockReq, {
+        messageId: 'msg1',
+        conversationId,
+        text: 'First message',
+        user: 'user123',
+      });
+
+      await saveMessage(mockReq, {
+        messageId: 'msg2',
+        conversationId,
+        text: 'Second message',
+        user: 'user123',
+      });
+
+      const messages = await getMessages({ conversationId });
+      expect(messages).toHaveLength(2);
+      expect(messages[0].text).toBe('First message');
+      expect(messages[1].text).toBe('Second message');
    });
  });

  describe('deleteMessages', () => {
    it('should delete messages with the correct filter', async () => {
+      // Save some messages for different users
+      await saveMessage(mockReq, mockMessageData);
+      await saveMessage(
+        { user: { id: 'user456' } },
+        {
+          messageId: 'msg456',
+          conversationId: uuidv4(),
+          text: 'Other user message',
+          user: 'user456',
+        },
+      );
+
      await deleteMessages({ user: 'user123' });
-      expect(mockSchema.deleteMany).toHaveBeenCalledWith({ user: 'user123' });
+
+      // Verify only user123's messages were deleted
+      const user123Messages = await Message.find({ user: 'user123' });
+      const user456Messages = await Message.find({ user: 'user456' });
+
+      expect(user123Messages).toHaveLength(0);
+      expect(user456Messages).toHaveLength(1);
    });
  });

  describe('Conversation Hijacking Prevention', () => {
-    it('should not allow editing a message in another user\'s conversation', async () => {
+    it("should not allow editing a message in another user's conversation", async () => {
      const attackerReq = { user: { id: 'attacker123' } };
-      const victimConversationId = 'victim-convo-123';
+      const victimConversationId = uuidv4();
      const victimMessageId = 'victim-msg-123';

-      mockSchema.findOneAndUpdate.mockResolvedValue(null);
+      // First, save a message as the victim (but we'll try to edit as attacker)
+      const victimReq = { user: { id: 'victim123' } };
+      await saveMessage(victimReq, {
+        messageId: victimMessageId,
+        conversationId: victimConversationId,
+        text: 'Victim message',
+        user: 'victim123',
+      });

+      // Attacker tries to edit the victim's message
      await expect(
        updateMessage(attackerReq, {
          messageId: victimMessageId,
@@ -168,71 +236,82 @@ describe('Message Operations', () => {
        }),
      ).rejects.toThrow('Message not found or user not authorized.');

-      expect(mockSchema.findOneAndUpdate).toHaveBeenCalledWith(
-        { messageId: victimMessageId, user: 'attacker123' },
-        expect.anything(),
-        expect.anything(),
-      );
+      // Verify the original message is unchanged
+      const originalMessage = await Message.findOne({
+        messageId: victimMessageId,
+        user: 'victim123',
+      });
+      expect(originalMessage.text).toBe('Victim message');
    });

-    it('should not allow deleting messages from another user\'s conversation', async () => {
+    it("should not allow deleting messages from another user's conversation", async () => {
      const attackerReq = { user: { id: 'attacker123' } };
-      const victimConversationId = 'victim-convo-123';
+      const victimConversationId = uuidv4();
      const victimMessageId = 'victim-msg-123';

-      mockSchema.findOne().lean.mockResolvedValueOnce(null); // Simulating message not found for this user
+      // Save a message as the victim
+      const victimReq = { user: { id: 'victim123' } };
+      await saveMessage(victimReq, {
+        messageId: victimMessageId,
+        conversationId: victimConversationId,
+        text: 'Victim message',
+        user: 'victim123',
+      });
+
+      // Attacker tries to delete from victim's conversation
      const result = await deleteMessagesSince(attackerReq, {
        messageId: victimMessageId,
        conversationId: victimConversationId,
      });

      expect(result).toBeUndefined();
-      expect(mockSchema.findOne).toHaveBeenCalledWith({
+
+      // Verify the victim's message still exists
+      const victimMessage = await Message.findOne({
        messageId: victimMessageId,
-        user: 'attacker123',
+        user: 'victim123',
      });
+      expect(victimMessage).toBeTruthy();
+      expect(victimMessage.text).toBe('Victim message');
    });

-    it('should not allow inserting a new message into another user\'s conversation', async () => {
+    it("should not allow inserting a new message into another user's conversation", async () => {
      const attackerReq = { user: { id: 'attacker123' } };
-      const victimConversationId = uuidv4(); // Use a valid UUID
+      const victimConversationId = uuidv4();

-      await expect(
-        saveMessage(attackerReq, {
-          conversationId: victimConversationId,
-          text: 'Inserted malicious message',
-          messageId: 'new-msg-123',
-        }),
-      ).resolves.not.toThrow(); // It should not throw an error
+      // Attacker tries to save a message - this should succeed but with attacker's user ID
+      const result = await saveMessage(attackerReq, {
+        conversationId: victimConversationId,
+        text: 'Inserted malicious message',
+        messageId: 'new-msg-123',
+        user: 'attacker123',
+      });

-      // Check that the message was saved with the attacker's user ID
-      expect(mockSchema.findOneAndUpdate).toHaveBeenCalledWith(
-        { messageId: 'new-msg-123', user: 'attacker123' },
-        expect.objectContaining({
-          user: 'attacker123',
-          conversationId: victimConversationId,
-        }),
-        expect.anything(),
-      );
+      expect(result).toBeTruthy();
+      expect(result.user).toBe('attacker123');
+
+      // Verify the message was saved with the attacker's user ID, not as an anonymous message
+      const savedMessage = await Message.findOne({ messageId: 'new-msg-123' });
+      expect(savedMessage.user).toBe('attacker123');
+      expect(savedMessage.conversationId).toBe(victimConversationId);
    });

    it('should allow retrieving messages from any conversation', async () => {
-      const victimConversationId = 'victim-convo-123';
+      const victimConversationId = uuidv4();

-      await getMessages({ conversationId: victimConversationId });
-
-      expect(mockSchema.find).toHaveBeenCalledWith({
+      // Save a message in the victim's conversation
+      const victimReq = { user: { id: 'victim123' } };
+      await saveMessage(victimReq, {
+        messageId: 'victim-msg',
        conversationId: victimConversationId,
+        text: 'Victim message',
+        user: 'victim123',
      });

-      mockSchema.find.mockReturnValueOnce({
-        select: jest.fn().mockReturnThis(),
-        sort: jest.fn().mockReturnThis(),
-        lean: jest.fn().mockResolvedValue([{ text: 'Test message' }]),
-      });
-
-      const result = await getMessages({ conversationId: victimConversationId });
-      expect(result).toEqual([{ text: 'Test message' }]);
+      // Anyone should be able to retrieve messages by conversation ID
+      const messages = await getMessages({ conversationId: victimConversationId });
+      expect(messages).toHaveLength(1);
+      expect(messages[0].text).toBe('Victim message');
    });
  });
 });
--- a/api/models/Preset.js
+++ b/api/models/Preset.js
@@ -1,5 +1,5 @@
-const Preset = require('./schema/presetSchema');
-const { logger } = require('~/config');
+const { logger } = require('@librechat/data-schemas');
+const { Preset } = require('~/db/models');

 const getPreset = async (user, presetId) => {
  try {
@@ -11,7 +11,6 @@ const getPreset = async (user, presetId) => {
 };

 module.exports = {
-  Preset,
  getPreset,
  getPresets: async (user, filter) => {
    try {
--- a/api/models/Project.js
+++ b/api/models/Project.js
@@ -1,15 +1,12 @@
-const { model } = require('mongoose');
 const { GLOBAL_PROJECT_NAME } = require('librechat-data-provider').Constants;
-const projectSchema = require('~/models/schema/projectSchema');
-
-const Project = model('Project', projectSchema);
+const { Project } = require('~/db/models');

 /**
 * Retrieve a project by ID and convert the found project document to a plain object.
 *
 * @param {string} projectId - The ID of the project to find and return as a plain object.
 * @param {string|string[]} [fieldsToSelect] - The fields to include or exclude in the returned document.
- * @returns {Promise<MongoProject>} A plain object representing the project document, or `null` if no project is found.
+ * @returns {Promise<IMongoProject>} A plain object representing the project document, or `null` if no project is found.
 */
 const getProjectById = async function (projectId, fieldsToSelect = null) {
  const query = Project.findById(projectId);
@@ -27,7 +24,7 @@ const getProjectById = async function (projectId, fieldsToSelect = null) {
 *
 * @param {string} projectName - The name of the project to find or create.
 * @param {string|string[]} [fieldsToSelect] - The fields to include or exclude in the returned document.
- * @returns {Promise<MongoProject>} A plain object representing the project document.
+ * @returns {Promise<IMongoProject>} A plain object representing the project document.
 */
 const getProjectByName = async function (projectName, fieldsToSelect = null) {
  const query = { name: projectName };
@@ -47,7 +44,7 @@ const getProjectByName = async function (projectName, fieldsToSelect = null) {
 *
 * @param {string} projectId - The ID of the project to update.
 * @param {string[]} promptGroupIds - The array of prompt group IDs to add to the project.
- * @returns {Promise<MongoProject>} The updated project document.
+ * @returns {Promise<IMongoProject>} The updated project document.
 */
 const addGroupIdsToProject = async function (projectId, promptGroupIds) {
  return await Project.findByIdAndUpdate(
@@ -62,7 +59,7 @@ const addGroupIdsToProject = async function (projectId, promptGroupIds) {
 *
 * @param {string} projectId - The ID of the project to update.
 * @param {string[]} promptGroupIds - The array of prompt group IDs to remove from the project.
- * @returns {Promise<MongoProject>} The updated project document.
+ * @returns {Promise<IMongoProject>} The updated project document.
 */
 const removeGroupIdsFromProject = async function (projectId, promptGroupIds) {
  return await Project.findByIdAndUpdate(
@@ -87,7 +84,7 @@ const removeGroupFromAllProjects = async (promptGroupId) => {
 *
 * @param {string} projectId - The ID of the project to update.
 * @param {string[]} agentIds - The array of agent IDs to add to the project.
- * @returns {Promise<MongoProject>} The updated project document.
+ * @returns {Promise<IMongoProject>} The updated project document.
 */
 const addAgentIdsToProject = async function (projectId, agentIds) {
  return await Project.findByIdAndUpdate(
@@ -102,7 +99,7 @@ const addAgentIdsToProject = async function (projectId, agentIds) {
 *
 * @param {string} projectId - The ID of the project to update.
 * @param {string[]} agentIds - The array of agent IDs to remove from the project.
- * @returns {Promise<MongoProject>} The updated project document.
+ * @returns {Promise<IMongoProject>} The updated project document.
 */
 const removeAgentIdsFromProject = async function (projectId, agentIds) {
  return await Project.findByIdAndUpdate(
--- a/api/models/Prompt.js
+++ b/api/models/Prompt.js
@@ -1,4 +1,5 @@
 const { ObjectId } = require('mongodb');
+const { logger } = require('@librechat/data-schemas');
 const { SystemRoles, SystemCategories, Constants } = require('librechat-data-provider');
 const {
  getProjectByName,
@@ -6,9 +7,8 @@ const {
  removeGroupIdsFromProject,
  removeGroupFromAllProjects,
 } = require('./Project');
-const { Prompt, PromptGroup } = require('./schema/promptSchema');
+const { PromptGroup, Prompt } = require('~/db/models');
 const { escapeRegExp } = require('~/server/utils');
-const { logger } = require('~/config');

 /**
 * Create a pipeline for the aggregation to get prompt groups
--- a/api/models/Role.js
+++ b/api/models/Role.js
@@ -3,27 +3,25 @@ const {
  SystemRoles,
  roleDefaults,
  PermissionTypes,
+  permissionsSchema,
  removeNullishValues,
-  agentPermissionsSchema,
-  promptPermissionsSchema,
-  bookmarkPermissionsSchema,
-  multiConvoPermissionsSchema,
 } = require('librechat-data-provider');
+const { logger } = require('@librechat/data-schemas');
 const getLogStores = require('~/cache/getLogStores');
-const Role = require('~/models/schema/roleSchema');
-const { logger } = require('~/config');
+const { Role } = require('~/db/models');

 /**
 * Retrieve a role by name and convert the found role document to a plain object.
- * If the role with the given name doesn't exist and the name is a system defined role, create it and return the lean version.
+ * If the role with the given name doesn't exist and the name is a system defined role,
+ * create it and return the lean version.
 *
 * @param {string} roleName - The name of the role to find or create.
 * @param {string|string[]} [fieldsToSelect] - The fields to include or exclude in the returned document.
 * @returns {Promise<Object>} A plain object representing the role document.
 */
 const getRoleByName = async function (roleName, fieldsToSelect = null) {
+  const cache = getLogStores(CacheKeys.ROLES);
  try {
-    const cache = getLogStores(CacheKeys.ROLES);
    const cachedRole = await cache.get(roleName);
    if (cachedRole) {
      return cachedRole;
@@ -35,8 +33,7 @@ const getRoleByName = async function (roleName, fieldsToSelect = null) {
    let role = await query.lean().exec();

    if (!role && SystemRoles[roleName]) {
-      role = roleDefaults[roleName];
-      role = await new Role(role).save();
+      role = await new Role(roleDefaults[roleName]).save();
      await cache.set(roleName, role);
      return role.toObject();
    }
@@ -55,8 +52,8 @@ const getRoleByName = async function (roleName, fieldsToSelect = null) {
 * @returns {Promise<TRole>} Updated role document.
 */
 const updateRoleByName = async function (roleName, updates) {
+  const cache = getLogStores(CacheKeys.ROLES);
  try {
-    const cache = getLogStores(CacheKeys.ROLES);
    const role = await Role.findOneAndUpdate(
      { name: roleName },
      { $set: updates },
@@ -72,27 +69,20 @@ const updateRoleByName = async function (roleName, updates) {
  }
 };

-const permissionSchemas = {
-  [PermissionTypes.AGENTS]: agentPermissionsSchema,
-  [PermissionTypes.PROMPTS]: promptPermissionsSchema,
-  [PermissionTypes.BOOKMARKS]: bookmarkPermissionsSchema,
-  [PermissionTypes.MULTI_CONVO]: multiConvoPermissionsSchema,
-};
-
 /**
 * Updates access permissions for a specific role and multiple permission types.
- * @param {SystemRoles} roleName - The role to update.
+ * @param {string} roleName - The role to update.
 * @param {Object.<PermissionTypes, Object.<Permissions, boolean>>} permissionsUpdate - Permissions to update and their values.
 */
 async function updateAccessPermissions(roleName, permissionsUpdate) {
+  // Filter and clean the permission updates based on our schema definition.
  const updates = {};
  for (const [permissionType, permissions] of Object.entries(permissionsUpdate)) {
-    if (permissionSchemas[permissionType]) {
+    if (permissionsSchema.shape && permissionsSchema.shape[permissionType]) {
      updates[permissionType] = removeNullishValues(permissions);
    }
  }
-
-  if (Object.keys(updates).length === 0) {
+  if (!Object.keys(updates).length) {
    return;
  }

@@ -102,26 +92,75 @@ async function updateAccessPermissions(roleName, permissionsUpdate) {
      return;
    }

-    const updatedPermissions = {};
+    const currentPermissions = role.permissions || {};
+    const updatedPermissions = { ...currentPermissions };
    let hasChanges = false;

+    const unsetFields = {};
+    const permissionTypes = Object.keys(permissionsSchema.shape || {});
+    for (const permType of permissionTypes) {
+      if (role[permType] && typeof role[permType] === 'object') {
+        logger.info(
+          `Migrating '${roleName}' role from old schema: found '${permType}' at top level`,
+        );
+
+        updatedPermissions[permType] = {
+          ...updatedPermissions[permType],
+          ...role[permType],
+        };
+
+        unsetFields[permType] = 1;
+        hasChanges = true;
+      }
+    }
+
+    // Process the current updates
    for (const [permissionType, permissions] of Object.entries(updates)) {
-      const currentPermissions = role[permissionType] || {};
-      updatedPermissions[permissionType] = { ...currentPermissions };
+      const currentTypePermissions = currentPermissions[permissionType] || {};
+      updatedPermissions[permissionType] = { ...currentTypePermissions };

      for (const [permission, value] of Object.entries(permissions)) {
-        if (currentPermissions[permission] !== value) {
+        if (currentTypePermissions[permission] !== value) {
          updatedPermissions[permissionType][permission] = value;
          hasChanges = true;
          logger.info(
-            `Updating '${roleName}' role ${permissionType} '${permission}' permission from ${currentPermissions[permission]} to: ${value}`,
+            `Updating '${roleName}' role permission '${permissionType}' '${permission}' from ${currentTypePermissions[permission]} to: ${value}`,
          );
        }
      }
    }

    if (hasChanges) {
-      await updateRoleByName(roleName, updatedPermissions);
+      const updateObj = { permissions: updatedPermissions };
+
+      if (Object.keys(unsetFields).length > 0) {
+        logger.info(
+          `Unsetting old schema fields for '${roleName}' role: ${Object.keys(unsetFields).join(', ')}`,
+        );
+
+        try {
+          await Role.updateOne(
+            { name: roleName },
+            {
+              $set: updateObj,
+              $unset: unsetFields,
+            },
+          );
+
+          const cache = getLogStores(CacheKeys.ROLES);
+          const updatedRole = await Role.findOne({ name: roleName }).select('-__v').lean().exec();
+          await cache.set(roleName, updatedRole);
+
+          logger.info(`Updated role '${roleName}' and removed old schema fields`);
+        } catch (updateError) {
+          logger.error(`Error during role migration update: ${updateError.message}`);
+          throw updateError;
+        }
+      } else {
+        // Standard update if no migration needed
+        await updateRoleByName(roleName, updateObj);
+      }
+
      logger.info(`Updated '${roleName}' role permissions`);
    } else {
      logger.info(`No changes needed for '${roleName}' role permissions`);
@@ -132,40 +171,87 @@ async function updateAccessPermissions(roleName, permissionsUpdate) {
 }

 /**
- * Initialize default roles in the system.
- * Creates the default roles (ADMIN, USER) if they don't exist in the database.
- * Updates existing roles with new permission types if they're missing.
+ * Migrates roles from old schema to new schema structure.
+ * This can be called directly to fix existing roles.
 *
- * @returns {Promise<void>}
+ * @param {string} [roleName] - Optional specific role to migrate. If not provided, migrates all roles.
+ * @returns {Promise<number>} Number of roles migrated.
 */
-const initializeRoles = async function () {
-  const defaultRoles = [SystemRoles.ADMIN, SystemRoles.USER];
-
-  for (const roleName of defaultRoles) {
-    let role = await Role.findOne({ name: roleName });
-
-    if (!role) {
-      // Create new role if it doesn't exist
-      role = new Role(roleDefaults[roleName]);
+const migrateRoleSchema = async function (roleName) {
+  try {
+    // Get roles to migrate
+    let roles;
+    if (roleName) {
+      const role = await Role.findOne({ name: roleName });
+      roles = role ? [role] : [];
    } else {
-      // Add missing permission types
-      let isUpdated = false;
-      for (const permType of Object.values(PermissionTypes)) {
-        if (!role[permType]) {
-          role[permType] = roleDefaults[roleName][permType];
-          isUpdated = true;
+      roles = await Role.find({});
+    }
+
+    logger.info(`Migrating ${roles.length} roles to new schema structure`);
+    let migratedCount = 0;
+
+    for (const role of roles) {
+      const permissionTypes = Object.keys(permissionsSchema.shape || {});
+      const unsetFields = {};
+      let hasOldSchema = false;
+
+      // Check for old schema fields
+      for (const permType of permissionTypes) {
+        if (role[permType] && typeof role[permType] === 'object') {
+          hasOldSchema = true;
+
+          // Ensure permissions object exists
+          role.permissions = role.permissions || {};
+
+          // Migrate permissions from old location to new
+          role.permissions[permType] = {
+            ...role.permissions[permType],
+            ...role[permType],
+          };
+
+          // Mark field for removal
+          unsetFields[permType] = 1;
        }
      }
-      if (isUpdated) {
-        await role.save();
+
+      if (hasOldSchema) {
+        try {
+          logger.info(`Migrating role '${role.name}' from old schema structure`);
+
+          // Simple update operation
+          await Role.updateOne(
+            { _id: role._id },
+            {
+              $set: { permissions: role.permissions },
+              $unset: unsetFields,
+            },
+          );
+
+          // Refresh cache
+          const cache = getLogStores(CacheKeys.ROLES);
+          const updatedRole = await Role.findById(role._id).lean().exec();
+          await cache.set(role.name, updatedRole);
+
+          migratedCount++;
+          logger.info(`Migrated role '${role.name}'`);
+        } catch (error) {
+          logger.error(`Failed to migrate role '${role.name}': ${error.message}`);
+        }
      }
    }
-    await role.save();
+
+    logger.info(`Migration complete: ${migratedCount} roles migrated`);
+    return migratedCount;
+  } catch (error) {
+    logger.error(`Role schema migration failed: ${error.message}`);
+    throw error;
  }
 };
+
 module.exports = {
  getRoleByName,
-  initializeRoles,
  updateRoleByName,
+  migrateRoleSchema,
  updateAccessPermissions,
 };
--- a/Show More
+++ b/Show More