chore: Remove ClosePlugin from Vite config

chore: close plugin
chore: install/declare rollup at latest, update vite precache
2024-08-28 13:00:01 -04:00 · 2024-08-28 11:33:23 -04:00 · 2024-08-28 11:15:39 -04:00 · 2024-08-27 17:03:16 -04:00 · 2024-08-27 15:01:41 -04:00 · 2024-08-27 14:49:43 -04:00
1518 changed files with 196684 additions and 27897 deletions
--- a/.devcontainer/Dockerfile
+++ b/.devcontainer/Dockerfile
@@ -0,0 +1,5 @@
+FROM node:18-bullseye
+
+RUN useradd -m -s /bin/bash vscode
+RUN mkdir -p /workspaces && chown -R vscode:vscode /workspaces
+WORKDIR /workspaces
--- a/.devcontainer/devcontainer.json
+++ b/.devcontainer/devcontainer.json
@@ -1,58 +1,18 @@
-// {
-// 	"name": "LibreChat_dev",
-// 	// Update the 'dockerComposeFile' list if you have more compose files or use different names.
-// 	"dockerComposeFile": "docker-compose.yml",
-// 	// The 'service' property is the name of the service for the container that VS Code should
-// 	// use. Update this value and .devcontainer/docker-compose.yml to the real service name.
-// 	"service": "librechat",
-// 	// The 'workspaceFolder' property is the path VS Code should open by default when
-// 	// connected. Corresponds to a volume mount in .devcontainer/docker-compose.yml
-// 	"workspaceFolder": "/workspace"
-//     //,
-// 	// // Set *default* container specific settings.json values on container create.
-// 	// "settings": {},
-// 	// // Add the IDs of extensions you want installed when the container is created.
-// 	// "extensions": [],
-// 	// Uncomment the next line if you want to keep your containers running after VS Code shuts down.
-// 	// "shutdownAction": "none",
-// 	// Uncomment the next line to use 'postCreateCommand' to run commands after the container is created.
-// 	// "postCreateCommand": "uname -a",
-// 	// Comment out to connect as root instead. To add a non-root user, see: https://aka.ms/vscode-remote/containers/non-root.
-// 	// "remoteUser": "vscode"
-// }
 {
-    // "name": "LibreChat_dev",
-    "dockerComposeFile": "docker-compose.yml",
-    "service": "app",
-    // "image": "node:19-alpine",
-    // "workspaceFolder": "/workspaces",
-    "workspaceFolder": "/workspace",
-    // Set *default* container specific settings.json values on container create.
-    // "overrideCommand": true,
-    "customizations": {
-        "vscode": {
-            "extensions": [],
-            "settings": {
-                "terminal.integrated.profiles.linux": {
-                    "bash": null
-                }
-            }
+  "dockerComposeFile": "docker-compose.yml",
+  "service": "app",
+  "workspaceFolder": "/workspaces",
+  "customizations": {
+    "vscode": {
+      "extensions": [],
+      "settings": {
+        "terminal.integrated.profiles.linux": {
+          "bash": null
        }
-    },
-    "postCreateCommand": "",
-    // "workspaceMount": "src=${localWorkspaceFolder},dst=/code,type=bind,consistency=cached"
-
-    // "runArgs": [
-    //     "--cap-add=SYS_PTRACE", "--security-opt", "seccomp=unconfined",
-    //     "-v", "/tmp/.X11-unix:/tmp/.X11-unix",
-    //     "-v", "${env:XAUTHORITY}:/root/.Xauthority:rw",
-    //     "-v", "/home/${env:USER}/.cdh:/root/.cdh",
-    //     "-e", "DISPLAY=${env:DISPLAY}",
-    //     "--name=tgw_assistant_backend_dev",
-    //     "--network=host"
-    // ],
-    // "settings": {
-    //     "terminal.integrated.shell.linux": "/bin/bash"
-    // },
-    "features": {"ghcr.io/devcontainers/features/git:1": {}}
+      }
+    }
+  },
+  "postCreateCommand": "",
+  "features": { "ghcr.io/devcontainers/features/git:1": {} },
+  "remoteUser": "vscode"
 }
--- a/.devcontainer/docker-compose.yml
+++ b/.devcontainer/docker-compose.yml
@@ -1,17 +1,11 @@
-version: '3.4'
+version: "3.8"

 services:
  app:
-    # container_name: LibreChat_dev
-    image: node:19-bullseye
-    # Using a Dockerfile is optional, but included for completeness.
-    # build: 
-    #   context: .
-    #   dockerfile: Dockerfile
-    #   # [Optional] You can use build args to set options. e.g. 'VARIANT' below affects the image in the Dockerfile
-    #   args: 
-    #     VARIANT: buster
-    # network_mode: "host"
+    build:
+      context: ..
+      dockerfile: .devcontainer/Dockerfile
+    # restart: always
    links:
      - mongodb
      - meilisearch
@@ -21,17 +15,16 @@ services:
    - "host.docker.internal:host-gateway"

    volumes:
-    #   # This is where VS Code should expect to find your project's source code and the value of "workspaceFolder" in .devcontainer/devcontainer.json
-      - ..:/workspace:cached
-    #   # - /app/client/node_modules
-    #   # - ./api:/app/api
-    #   # - ./.env:/app/.env
-    #   # - ./.env.development:/app/.env.development
-    #   # - ./.env.production:/app/.env.production
-    #   # - /app/api/node_modules
-
-    #   # Uncomment the next line to use Docker from inside the container. See https://aka.ms/vscode-remote/samples/docker-from-docker-compose for details.
-    #   # - /var/run/docker.sock:/var/run/docker.sock 
+      # This is where VS Code should expect to find your project's source code and the value of "workspaceFolder" in .devcontainer/devcontainer.json
+      - ..:/workspaces:cached
+      # Uncomment the next line to use Docker from inside the container. See https://aka.ms/vscode-remote/samples/docker-from-docker-compose for details.
+      # - /var/run/docker.sock:/var/run/docker.sock 
+    environment:
+      - HOST=0.0.0.0
+      - MONGO_URI=mongodb://mongodb:27017/LibreChat
+      # - CHATGPT_REVERSE_PROXY=http://host.docker.internal:8080/api/conversation # if you are hosting your own chatgpt reverse proxy with docker
+      # - OPENAI_REVERSE_PROXY=http://host.docker.internal:8070/v1/chat/completions # if you are hosting your own chatgpt reverse proxy with docker
+      - MEILI_HOST=http://meilisearch:7700

    # Runs app on the same network as the service container, allows "forwardPorts" in devcontainer.json function.
    # network_mode: service:another-service
@@ -39,45 +32,34 @@ services:
    # Use "forwardPorts" in **devcontainer.json** to forward an app port locally. 
    # (Adding the "ports" property to this file will not forward from a Codespace.)

-    # Uncomment the next line to use a non-root user for all processes - See https://aka.ms/vscode-remote/containers/non-root for details.
-    # user: vscode
-
-    # Uncomment the next four lines if you will use a ptrace-based debugger like C++, Go, and Rust.
-    # cap_add:
-    #   - SYS_PTRACE
-    # security_opt:
-    #   - seccomp:unconfined
+    # Use a non-root user for all processes - See https://aka.ms/vscode-remote/containers/non-root for details.
+    user: vscode

    # Overrides default command so things don't shut down after the process ends.
    command: /bin/sh -c "while sleep 1000; do :; done"  

  mongodb:
    container_name: chat-mongodb
-    # network_mode: "host"
    expose:
      - 27017
    # ports:
    #   - 27018:27017
    image: mongo
-  #   restart: always
+    # restart: always
    volumes:
      - ./data-node:/data/db
    command: mongod --noauth
  meilisearch:
    container_name: chat-meilisearch
-    image: getmeili/meilisearch:v1.0
-    # network_mode: "host"
+    image: getmeili/meilisearch:v1.5
+    # restart: always
    expose:
      - 7700
-    # ports:
-    #   - 7700:7700
-    # env_file:
-    #   - .env
+    # Uncomment this to access meilisearch from outside docker
+    # ports: 
+    #   - 7700:7700 # if exposing these ports, make sure your master key is not the default value
    environment:
-      - SEARCH=false
-      - MEILI_HOST=http://0.0.0.0:7700
-      - MEILI_HTTP_ADDR=0.0.0.0:7700
+      - MEILI_NO_ANALYTICS=true
      - MEILI_MASTER_KEY=5c71cf56d672d009e36070b5bc5e47b743535ae55c818ae3b735bb6ebfb4ba63
    volumes:
-      - ./meili_data:/meili_data
-  
+      - ./meili_data_v1.5:/meili_data
--- a/.dockerignore
+++ b/.dockerignore
@@ -1,5 +1,17 @@
+**/.circleci
+**/.editorconfig
+**/.dockerignore
+**/.git
+**/.DS_Store
+**/.vscode
 **/node_modules
-client/dist/images
+
+# Specific patterns to ignore
 data-node
-.env
-**/.env
+meili_data*
+librechat*
+Dockerfile*
+docs
+
+# Ignore all hidden files
+.*
--- a/.env.example
+++ b/.env.example
@@ -1,29 +1,303 @@
-##########################
-# Server configuration: 
-##########################
+#=====================================================================#
+#                       LibreChat Configuration                       #
+#=====================================================================#
+# Please refer to the reference documentation for assistance          #
+# with configuring your LibreChat environment.                        #
+#                                                                     #
+# https://www.librechat.ai/docs/configuration/dotenv                  #
+#=====================================================================#

-APP_TITLE=LibreChat
+#==================================================#
+#               Server Configuration               #
+#==================================================#

-# The server will listen to localhost:3080 by default. You can change the target IP as you want. 
-# If you want to make this server available externally, for example to share the server with others 
-# or expose this from a Docker container, set host to 0.0.0.0 or your external IP interface. 
-# Tips: Setting host to 0.0.0.0 means listening on all interfaces. It's not a real IP.  
-# Use localhost:port rather than 0.0.0.0:port to access the server. 
-# Set Node env to development if running in dev mode.
 HOST=localhost
 PORT=3080

-# Automated Moderation System
-# The Automated Moderation System uses a scoring mechanism to track user violations. As users commit actions
-# like excessive logins, registrations, or messaging, they accumulate violation scores. Upon reaching
-# a set threshold, the user and their IP are temporarily banned. This system ensures platform security
-# by monitoring and penalizing rapid or suspicious activities.
+MONGO_URI=mongodb://127.0.0.1:27017/LibreChat

-BAN_VIOLATIONS=true # Whether or not to enable banning users for violations (they will still be logged)
-BAN_DURATION=1000 * 60 * 60 * 2 # how long the user and associated IP are banned for
-BAN_INTERVAL=20 # a user will be banned everytime their score reaches/crosses over the interval threshold
+DOMAIN_CLIENT=http://localhost:3080
+DOMAIN_SERVER=http://localhost:3080

-# The score for each violation
+NO_INDEX=true
+
+#===============#
+# JSON Logging  #
+#===============#
+
+# Use when process console logs in cloud deployment like GCP/AWS
+CONSOLE_JSON=false
+
+#===============#
+# Debug Logging #
+#===============#
+
+DEBUG_LOGGING=true
+DEBUG_CONSOLE=false
+
+#=============#
+# Permissions #
+#=============#
+
+# UID=1000
+# GID=1000
+
+#===============#
+# Configuration #
+#===============#
+# Use an absolute path, a relative path, or a URL
+
+# CONFIG_PATH="/alternative/path/to/librechat.yaml"
+
+#===================================================#
+#                     Endpoints                     #
+#===================================================#
+
+# ENDPOINTS=openAI,assistants,azureOpenAI,bingAI,google,gptPlugins,anthropic
+
+PROXY=
+
+#===================================#
+# Known Endpoints - librechat.yaml  #
+#===================================#
+# https://www.librechat.ai/docs/configuration/librechat_yaml/ai_endpoints
+
+# ANYSCALE_API_KEY=
+# APIPIE_API_KEY=
+# COHERE_API_KEY=
+# DEEPSEEK_API_KEY=
+# DATABRICKS_API_KEY=
+# FIREWORKS_API_KEY=
+# GROQ_API_KEY=
+# HUGGINGFACE_TOKEN=
+# MISTRAL_API_KEY=
+# OPENROUTER_KEY=
+# PERPLEXITY_API_KEY=
+# SHUTTLEAI_API_KEY=
+# TOGETHERAI_API_KEY=
+# UNIFY_API_KEY=
+
+#============#
+# Anthropic  #
+#============#
+
+ANTHROPIC_API_KEY=user_provided
+# ANTHROPIC_MODELS=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_REVERSE_PROXY=
+
+#============#
+# Azure      #
+#============#
+
+# Note: these variables are DEPRECATED
+# Use the `librechat.yaml` configuration for `azureOpenAI` instead
+# You may also continue to use them if you opt out of using the `librechat.yaml` configuration
+
+# AZURE_OPENAI_DEFAULT_MODEL=gpt-3.5-turbo # Deprecated
+# AZURE_OPENAI_MODELS=gpt-3.5-turbo,gpt-4 # Deprecated
+# AZURE_USE_MODEL_AS_DEPLOYMENT_NAME=TRUE # Deprecated
+# AZURE_API_KEY= # Deprecated
+# AZURE_OPENAI_API_INSTANCE_NAME= # Deprecated
+# AZURE_OPENAI_API_DEPLOYMENT_NAME= # Deprecated
+# AZURE_OPENAI_API_VERSION= # Deprecated
+# AZURE_OPENAI_API_COMPLETIONS_DEPLOYMENT_NAME= # Deprecated
+# AZURE_OPENAI_API_EMBEDDINGS_DEPLOYMENT_NAME= # Deprecated
+# PLUGINS_USE_AZURE="true" # Deprecated
+
+#============#
+# BingAI     #
+#============#
+
+BINGAI_TOKEN=user_provided
+# BINGAI_HOST=https://cn.bing.com
+
+#============#
+# Google     #
+#============#
+
+GOOGLE_KEY=user_provided
+# GOOGLE_REVERSE_PROXY=
+
+# Gemini API (AI Studio)
+# GOOGLE_MODELS=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
+
+# 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_TITLE_MODEL=gemini-pro
+
+# Google Safety Settings
+# NOTE: These settings apply to both Vertex AI and Gemini API (AI Studio)
+#
+# For Vertex AI:
+# To use the BLOCK_NONE setting, you need either:
+# (a) Access through an allowlist via your Google account team, or
+# (b) Switch to monthly invoiced billing: https://cloud.google.com/billing/docs/how-to/invoiced-billing
+#
+# For Gemini API (AI Studio):
+# BLOCK_NONE is available by default, no special account requirements.
+#
+# Available options: BLOCK_NONE, BLOCK_ONLY_HIGH, BLOCK_MEDIUM_AND_ABOVE, BLOCK_LOW_AND_ABOVE
+#
+# GOOGLE_SAFETY_SEXUALLY_EXPLICIT=BLOCK_ONLY_HIGH
+# GOOGLE_SAFETY_HATE_SPEECH=BLOCK_ONLY_HIGH
+# GOOGLE_SAFETY_HARASSMENT=BLOCK_ONLY_HIGH
+# GOOGLE_SAFETY_DANGEROUS_CONTENT=BLOCK_ONLY_HIGH
+
+#============#
+# OpenAI     #
+#============#
+
+OPENAI_API_KEY=user_provided
+# OPENAI_MODELS=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
+
+DEBUG_OPENAI=false
+
+# TITLE_CONVO=false
+# OPENAI_TITLE_MODEL=gpt-3.5-turbo
+
+# OPENAI_SUMMARIZE=true
+# OPENAI_SUMMARY_MODEL=gpt-3.5-turbo
+
+# OPENAI_FORCE_PROMPT=true
+
+# OPENAI_REVERSE_PROXY=
+
+# OPENAI_ORGANIZATION=
+
+#====================#
+#   Assistants API   #
+#====================#
+
+ASSISTANTS_API_KEY=user_provided
+# ASSISTANTS_BASE_URL=
+# ASSISTANTS_MODELS=gpt-4o,gpt-4o-mini,gpt-3.5-turbo-0125,gpt-3.5-turbo-16k-0613,gpt-3.5-turbo-16k,gpt-3.5-turbo,gpt-4,gpt-4-0314,gpt-4-32k-0314,gpt-4-0613,gpt-3.5-turbo-0613,gpt-3.5-turbo-1106,gpt-4-0125-preview,gpt-4-turbo-preview,gpt-4-1106-preview
+
+#==========================#
+#   Azure Assistants API   #
+#==========================#
+
+# Note: You should map your credentials with custom variables according to your Azure OpenAI Configuration
+# The models for Azure Assistants are also determined by your Azure OpenAI configuration.
+
+# 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    #
+#============#
+
+# PLUGIN_MODELS=gpt-4o,gpt-4o-mini,gpt-4,gpt-4-turbo-preview,gpt-4-0125-preview,gpt-4-1106-preview,gpt-4-0613,gpt-3.5-turbo,gpt-3.5-turbo-0125,gpt-3.5-turbo-1106,gpt-3.5-turbo-0613
+
+DEBUG_PLUGINS=true
+
+CREDS_KEY=f34be427ebb29de8d88c107a71546019685ed8b241d8f2ed00c3df97ad2566f0
+CREDS_IV=e2341419ec3dd3d19b13a1a87fafcbfb
+
+# Azure AI Search
+#-----------------
+AZURE_AI_SEARCH_SERVICE_ENDPOINT=
+AZURE_AI_SEARCH_INDEX_NAME=
+AZURE_AI_SEARCH_API_KEY=
+
+AZURE_AI_SEARCH_API_VERSION=
+AZURE_AI_SEARCH_SEARCH_OPTION_QUERY_TYPE=
+AZURE_AI_SEARCH_SEARCH_OPTION_TOP=
+AZURE_AI_SEARCH_SEARCH_OPTION_SELECT=
+
+# DALL·E
+#----------------
+# DALLE_API_KEY=
+# DALLE3_API_KEY=
+# DALLE2_API_KEY=
+# DALLE3_SYSTEM_PROMPT=
+# DALLE2_SYSTEM_PROMPT=
+# DALLE_REVERSE_PROXY=
+# DALLE3_BASEURL=
+# DALLE2_BASEURL=
+
+# DALL·E (via Azure OpenAI)
+# Note: requires some of the variables above to be set
+#----------------
+# DALLE3_AZURE_API_VERSION=
+# DALLE2_AZURE_API_VERSION=
+
+# Google
+#-----------------
+GOOGLE_SEARCH_API_KEY=
+GOOGLE_CSE_ID=
+
+# SerpAPI
+#-----------------
+SERPAPI_API_KEY=
+
+# Stable Diffusion
+#-----------------
+SD_WEBUI_URL=http://host.docker.internal:7860
+
+# Tavily
+#-----------------
+TAVILY_API_KEY=
+
+# Traversaal
+#-----------------
+TRAVERSAAL_API_KEY=
+
+# WolframAlpha
+#-----------------
+WOLFRAM_APP_ID=
+
+# Zapier
+#-----------------
+ZAPIER_NLA_API_KEY=
+
+#==================================================#
+#                      Search                      #
+#==================================================#
+
+SEARCH=true
+MEILI_NO_ANALYTICS=true
+MEILI_HOST=http://0.0.0.0:7700
+MEILI_MASTER_KEY=DrhYf7zENyR6AlUCKmnz0eYASOQdl6zxH7s7MKFSfFCt
+
+#==================================================#
+#          Speech to Text & Text to Speech         #
+#==================================================#
+
+STT_API_KEY=
+TTS_API_KEY=
+
+#==================================================#
+#                        RAG                       #
+#==================================================#
+# More info: https://www.librechat.ai/docs/configuration/rag_api
+
+# RAG_OPENAI_BASEURL=
+# RAG_OPENAI_API_KEY=
+# EMBEDDINGS_PROVIDER=openai
+# EMBEDDINGS_MODEL=text-embedding-3-small
+
+#===================================================#
+#                    User System                    #
+#===================================================#
+
+#========================#
+# Moderation             #
+#========================#
+
+OPENAI_MODERATION=false
+OPENAI_MODERATION_API_KEY=
+# OPENAI_MODERATION_REVERSE_PROXY=
+
+BAN_VIOLATIONS=true
+BAN_DURATION=1000 * 60 * 60 * 2
+BAN_INTERVAL=20

 LOGIN_VIOLATION_SCORE=1
 REGISTRATION_VIOLATION_SCORE=1
@@ -31,323 +305,162 @@ CONCURRENT_VIOLATION_SCORE=1
 MESSAGE_VIOLATION_SCORE=1
 NON_BROWSER_VIOLATION_SCORE=20

-# Login and registration rate limiting.
+LOGIN_MAX=7
+LOGIN_WINDOW=5
+REGISTER_MAX=5
+REGISTER_WINDOW=60

-LOGIN_MAX=7 # The max amount of logins allowed per IP per LOGIN_WINDOW
-LOGIN_WINDOW=5 # in minutes, determines the window of time for LOGIN_MAX logins
-REGISTER_MAX=5 # The max amount of registrations allowed per IP per REGISTER_WINDOW
-REGISTER_WINDOW=60 # in minutes, determines the window of time for REGISTER_MAX registrations
+LIMIT_CONCURRENT_MESSAGES=true
+CONCURRENT_MESSAGE_MAX=2

-# Message rate limiting (per user & IP)
+LIMIT_MESSAGE_IP=true
+MESSAGE_IP_MAX=40
+MESSAGE_IP_WINDOW=1

-LIMIT_CONCURRENT_MESSAGES=true # Whether to limit the amount of messages a user can send per request
-CONCURRENT_MESSAGE_MAX=2 # The max amount of messages a user can send per request
+LIMIT_MESSAGE_USER=false
+MESSAGE_USER_MAX=40
+MESSAGE_USER_WINDOW=1

-LIMIT_MESSAGE_IP=true # Whether to limit the amount of messages an IP can send per MESSAGE_IP_WINDOW
-MESSAGE_IP_MAX=40 # The max amount of messages an IP can send per MESSAGE_IP_WINDOW
-MESSAGE_IP_WINDOW=1 # in minutes, determines the window of time for MESSAGE_IP_MAX messages
+ILLEGAL_MODEL_REQ_SCORE=5

-# Note: You can utilize both limiters, but default is to limit by IP only.
-LIMIT_MESSAGE_USER=false # Whether to limit the amount of messages an IP can send per MESSAGE_USER_WINDOW
-MESSAGE_USER_MAX=40 # The max amount of messages an IP can send per MESSAGE_USER_WINDOW
-MESSAGE_USER_WINDOW=1 # in minutes, determines the window of time for MESSAGE_USER_MAX messages
+#========================#
+# Balance                #
+#========================#

-# Change this to proxy any API request. 
-# It's useful if your machine has difficulty calling the original API server. 
-# PROXY=
+CHECK_BALANCE=false

-# Change this to your MongoDB URI if different. I recommend appending LibreChat. 
-MONGO_URI=mongodb://127.0.0.1:27018/LibreChat
+#========================#
+# Registration and Login #
+#========================#

-##########################
-# OpenAI Endpoint: 
-##########################
-
-# Access key from OpenAI platform. 
-# Leave it blank to disable this feature.
-# Set to "user_provided" to allow the user to provide their API key from the UI.
-OPENAI_API_KEY=user_provided
-
-# Identify the available models, separated by commas *without spaces*. 
-# The first will be default. 
-# Leave it blank to use internal settings. 
-# OPENAI_MODELS=gpt-3.5-turbo,gpt-3.5-turbo-16k,gpt-3.5-turbo-0301,text-davinci-003,gpt-4,gpt-4-0314,gpt-4-0613
-
-# Reverse proxy settings for OpenAI: 
-# https://github.com/waylaidwanderer/node-chatgpt-api#using-a-reverse-proxy 
-# OPENAI_REVERSE_PROXY=
-
-##########################
-# OpenRouter (overrides OpenAI and Plugins Endpoints): 
-##########################
-
-# OpenRouter is a legitimate proxy service to a multitude of LLMs, both closed and open source, including:
-# OpenAI models, Anthropic models, Meta's Llama models, pygmalionai/mythalion-13b
-# and many more open source models. Newer integrations are usually discounted, too!
-
-# Note: this overrides the OpenAI and Plugins Endpoints.
-# See ./docs/install/free_ai_apis.md for more info.
-
-# OPENROUTER_API_KEY=
-
-##########################
-# AZURE Endpoint:   
-##########################
-
-# To use Azure with this project, set the following variables. These will be used to build the API URL.
-# Chat completion: 
-# `https://{AZURE_OPENAI_API_INSTANCE_NAME}.openai.azure.com/openai/deployments/{AZURE_OPENAI_API_DEPLOYMENT_NAME}/chat/completions?api-version={AZURE_OPENAI_API_VERSION}`;
-# You should also consider changing the `OPENAI_MODELS` variable above to the models available in your instance/deployment.
-# Note: I've noticed that the Azure API is much faster than the OpenAI API, so the streaming looks almost instantaneous.
-# Note "AZURE_OPENAI_API_COMPLETIONS_DEPLOYMENT_NAME" and "AZURE_OPENAI_API_EMBEDDINGS_DEPLOYMENT_NAME" are optional but might be used in the future
-
-# AZURE_API_KEY=
-# AZURE_OPENAI_API_INSTANCE_NAME=
-# AZURE_OPENAI_API_DEPLOYMENT_NAME=
-# AZURE_OPENAI_API_VERSION=
-# AZURE_OPENAI_API_COMPLETIONS_DEPLOYMENT_NAME=
-# AZURE_OPENAI_API_EMBEDDINGS_DEPLOYMENT_NAME=
-
-# Identify the available models, separated by commas *without spaces*. 
-# The first will be default. 
-# Leave it blank to use internal settings. 
-AZURE_OPENAI_MODELS=gpt-3.5-turbo,gpt-4
-
-# To use Azure with the Plugins endpoint, you need the variables above, and uncomment the following variable:
-# NOTE: This may not work as expected and Azure OpenAI may not support OpenAI Functions yet
-# Omit/leave it commented to use the default OpenAI API
-
-# PLUGINS_USE_AZURE="true"
-
-##########################
-# ChatGPT Endpoint: 
-##########################
-
-# ChatGPT Browser Client (free but use at your own risk) 
-# Access token from https://chat.openai.com/api/auth/session 
-# Exposes your access token to `CHATGPT_REVERSE_PROXY`
-# Set to "user_provided" to allow the user to provide its token from the UI.
-# Leave it blank to disable this endpoint 
-CHATGPT_TOKEN=user_provided
-
-# Identify the available models, separated by commas. The first will be default. 
-# Leave it blank to use internal settings. 
-CHATGPT_MODELS=text-davinci-002-render-sha,gpt-4
-# NOTE: you can add gpt-4-plugins, gpt-4-code-interpreter, and gpt-4-browsing to the list above and use the models for these features;
-# however, the view/display portion of these features are not supported, but you can use the underlying models, which have higher token context
-# Also: text-davinci-002-render-paid is deprecated as of May 2023
-
-# Reverse proxy setting for OpenAI
-# https://github.com/waylaidwanderer/node-chatgpt-api#using-a-reverse-proxy
-# By default it will use the node-chatgpt-api recommended proxy, (it's a third party server)
-# CHATGPT_REVERSE_PROXY=<YOUR REVERSE PROXY>
-
-##########################
-# BingAI Endpoint:   
-##########################
-
-# Also used for Sydney and jailbreak
-# To get your Access token for Bing, login to https://www.bing.com
-# Use dev tools or an extension while logged into the site to copy the content of the _U cookie. 
-# If this fails, follow these instructions https://github.com/danny-avila/LibreChat/issues/370#issuecomment-1560382302 to provide the full cookie strings 
-# or check out our discord https://discord.com/channels/1086345563026489514/1143941308684177429
-# Set to "user_provided" to allow the user to provide its token from the UI.
-# Leave it blank to disable this endpoint. 
-BINGAI_TOKEN=user_provided
-
-# BingAI Host: 
-# Necessary for some people in different countries, e.g. China (https://cn.bing.com) 
-# Leave it blank to use default server. 
-# BINGAI_HOST=https://cn.bing.com 
-
-#############################
-# Plugins:
-#############################
-
-# Identify the available models, separated by commas *without spaces*. 
-# The first will be default. 
-# Leave it blank to use internal settings. 
-# PLUGIN_MODELS=gpt-3.5-turbo,gpt-3.5-turbo-16k,gpt-3.5-turbo-0301,gpt-4,gpt-4-0314,gpt-4-0613
-
-DEBUG_PLUGINS=true # Set to false or comment out to disable debug mode for plugins
-
-# For securely storing credentials, you need a fixed key and IV. You can set them here for prod and dev environments
-# If you don't set them, the app will crash on startup.
-# You need a 32-byte key (64 characters in hex) and 16-byte IV (32 characters in hex)
-# Use this replit to generate some quickly: https://replit.com/@daavila/crypto#index.js
-# Here are some examples (THESE ARE NOT SECURE!)
-CREDS_KEY=f34be427ebb29de8d88c107a71546019685ed8b241d8f2ed00c3df97ad2566f0
-CREDS_IV=e2341419ec3dd3d19b13a1a87fafcbfb
-
-# AI-Assisted Google Search  
-# This bot supports searching google for answers to your questions with assistance from GPT! 
-# See detailed instructions here: https://github.com/danny-avila/LibreChat/blob/main/docs/features/plugins/google_search.md
-GOOGLE_API_KEY=
-GOOGLE_CSE_ID=
-
-# StableDiffusion WebUI
-# This bot supports StableDiffusion WebUI, using it's API to generated requested images.
-# See detailed instructions here: https://github.com/danny-avila/LibreChat/blob/main/docs/features/plugins/stable_diffusion.md
-# Use "http://127.0.0.1:7860" with local install and "http://host.docker.internal:7860" for docker
-SD_WEBUI_URL=http://host.docker.internal:7860
-
-# Azure Cognitive Search
-# This plugin supports searching Azure Cognitive Search for answers to your questions.
-# See detailed instructions here: https://github.com/danny-avila/LibreChat/blob/main/docs/features/plugins/azure_cognitive_search.md
-AZURE_COGNITIVE_SEARCH_SERVICE_ENDPOINT=
-AZURE_COGNITIVE_SEARCH_INDEX_NAME=
-AZURE_COGNITIVE_SEARCH_API_KEY=
-
-AZURE_COGNITIVE_SEARCH_API_VERSION=
-AZURE_COGNITIVE_SEARCH_SEARCH_OPTION_QUERY_TYPE=
-AZURE_COGNITIVE_SEARCH_SEARCH_OPTION_TOP=
-AZURE_COGNITIVE_SEARCH_SEARCH_OPTION_SELECT=
-
-##########################
-# PaLM (Google) Endpoint: 
-##########################
-
-# Follow the instruction here to setup:
-# https://github.com/danny-avila/LibreChat/blob/main/docs/install/apis_and_tokens.md
-
-PALM_KEY=user_provided
-
-# In case you need a reverse proxy for this endpoint:
-# GOOGLE_REVERSE_PROXY= 
-
-##########################
-# Anthropic Endpoint: 
-##########################
-# Access key from https://console.anthropic.com/
-# Leave it blank to disable this feature.
-# Set to "user_provided" to allow the user to provide their API key from the UI.
-# Note that access to claude-1 may potentially become unavailable with the release of claude-2.
-ANTHROPIC_API_KEY=user_provided
-ANTHROPIC_MODELS=claude-1,claude-instant-1,claude-2
-
-##########################
-# Proxy: To be Used by all endpoints
-##########################
-
-PROXY=
-
-##########################
-# Search: 
-##########################
-
-# ENABLING SEARCH MESSAGES/CONVOS 
-# Requires the installation of the free self-hosted Meilisearch or a paid Remote Plan (Remote not tested) 
-# The easiest setup for this is through docker-compose, which takes care of it for you. 
-SEARCH=true
-
-# HIGHLY RECOMMENDED: Disable anonymized telemetry analytics for MeiliSearch for absolute privacy.
-MEILI_NO_ANALYTICS=true
-
-# REQUIRED FOR SEARCH: MeiliSearch Host, mainly for the API server to connect to the search server. 
-# Replace '0.0.0.0' with 'meilisearch' if serving MeiliSearch with docker-compose. 
-MEILI_HOST=http://0.0.0.0:7700
-
-# REQUIRED FOR SEARCH: MeiliSearch HTTP Address, mainly for docker-compose to expose the search server. 
-# Replace '0.0.0.0' with 'meilisearch' if serving MeiliSearch with docker-compose. 
-MEILI_HTTP_ADDR=0.0.0.0:7700
-
-# REQUIRED FOR SEARCH: In production env., a secure key is needed. You can generate your own. 
-# This master key must be at least 16 bytes, composed of valid UTF-8 characters. 
-# MeiliSearch will throw an error and refuse to launch if no master key is provided, 
-# or if it is under 16 bytes. MeiliSearch will suggest a secure autogenerated master key.
-# Using docker, it seems recognized as production so use a secure key.
-# This is a ready made secure key for docker-compose, you can replace it with your own.
-MEILI_MASTER_KEY=DrhYf7zENyR6AlUCKmnz0eYASOQdl6zxH7s7MKFSfFCt
-
-##########################
-# User System:
-##########################
-
-# Allow Public Registration
+ALLOW_EMAIL_LOGIN=true
 ALLOW_REGISTRATION=true
-
-# Allow Social Registration
 ALLOW_SOCIAL_LOGIN=false
-
-# Allow Social Registration (WORKS ONLY for Google, Github, Discord)
 ALLOW_SOCIAL_REGISTRATION=false
+ALLOW_PASSWORD_RESET=false
+# ALLOW_ACCOUNT_DELETION=true # note: enabled by default if omitted/commented out
+ALLOW_UNVERIFIED_EMAIL_LOGIN=true
+
+SESSION_EXPIRY=1000 * 60 * 15
+REFRESH_TOKEN_EXPIRY=(1000 * 60 * 60 * 24) * 7

-# JWT Secrets
-# You should use secure values. The examples given are 32-byte keys (64 characters in hex)
-# Use this replit to generate some quickly: https://replit.com/@daavila/crypto#index.js
 JWT_SECRET=16f8c0ef4a5d391b26034086c628469d3f9f497f08163ab9b40137092f2909ef
 JWT_REFRESH_SECRET=eaa5191f2914e30b9387fd84e254e4ba6fc51b4654968a9b0803b456a54b8418

-# Google:
-# Add your Google Client ID and Secret here, you must register an app with Google Cloud to get these values
-# https://cloud.google.com/
-GOOGLE_CLIENT_ID=
-GOOGLE_CLIENT_SECRET=
-GOOGLE_CALLBACK_URL=/oauth/google/callback
+# Discord
+DISCORD_CLIENT_ID=
+DISCORD_CLIENT_SECRET=
+DISCORD_CALLBACK_URL=/oauth/discord/callback

-# Facebook:
-# Add your Facebook Client ID and Secret here, you must register an app with Facebook to get these values
-# https://developers.facebook.com/
+# Facebook
 FACEBOOK_CLIENT_ID=
 FACEBOOK_CLIENT_SECRET=
 FACEBOOK_CALLBACK_URL=/oauth/facebook/callback

-# OpenID:
-# See OpenID provider to get the below values
-# Create random string for OPENID_SESSION_SECRET
-# For Azure AD
-#   ISSUER:  https://login.microsoftonline.com/(tenant id)/v2.0/
-#   SCOPE:   openid profile email
+# GitHub
+GITHUB_CLIENT_ID=
+GITHUB_CLIENT_SECRET=
+GITHUB_CALLBACK_URL=/oauth/github/callback
+
+# Google
+GOOGLE_CLIENT_ID=
+GOOGLE_CLIENT_SECRET=
+GOOGLE_CALLBACK_URL=/oauth/google/callback
+
+# OpenID
 OPENID_CLIENT_ID=
 OPENID_CLIENT_SECRET=
 OPENID_ISSUER=
 OPENID_SESSION_SECRET=
 OPENID_SCOPE="openid profile email"
 OPENID_CALLBACK_URL=/oauth/openid/callback
-# If LABEL and URL are left empty, then the default OpenID label and logo are used.
+OPENID_REQUIRED_ROLE=
+OPENID_REQUIRED_ROLE_TOKEN_KIND=
+OPENID_REQUIRED_ROLE_PARAMETER_PATH=
+
 OPENID_BUTTON_LABEL=
 OPENID_IMAGE_URL=

-# Set the expiration delay for the secure cookie with the JWT token
-# Recommend session expiry to be 15 minutes
-# Delay is in millisecond e.g. 7 days is 1000*60*60*24*7
-SESSION_EXPIRY=1000 * 60 * 15
-REFRESH_TOKEN_EXPIRY=(1000 * 60 * 60 * 24) * 7
+# LDAP
+LDAP_URL=
+LDAP_BIND_DN=
+LDAP_BIND_CREDENTIALS=
+LDAP_USER_SEARCH_BASE=
+LDAP_SEARCH_FILTER=mail={{username}}
+LDAP_CA_CERT_PATH=
+# LDAP_TLS_REJECT_UNAUTHORIZED=
+# LDAP_LOGIN_USES_USERNAME=true
+# LDAP_ID=
+# LDAP_USERNAME=
+# LDAP_FULL_NAME=

-# Github:
-# Get the Client ID and Secret from your Discord Application
-# Add your Discord Client ID and Client Secret here:
+#========================#
+# Email Password Reset   #
+#========================#

-GITHUB_CLIENT_ID=your_client_id
-GITHUB_CLIENT_SECRET=your_client_secret
-GITHUB_CALLBACK_URL=/oauth/github/callback # this should be the same for everyone
+EMAIL_SERVICE=
+EMAIL_HOST=
+EMAIL_PORT=25
+EMAIL_ENCRYPTION=
+EMAIL_ENCRYPTION_HOSTNAME=
+EMAIL_ALLOW_SELFSIGNED=
+EMAIL_USERNAME=
+EMAIL_PASSWORD=
+EMAIL_FROM_NAME=
+EMAIL_FROM=noreply@librechat.ai

-# Discord:
-# Get the Client ID and Secret from your Discord Application
-# Add your Github Client ID and Client Secret here:
+#========================#
+# Firebase CDN           #
+#========================#

-DISCORD_CLIENT_ID=your_client_id
-DISCORD_CLIENT_SECRET=your_client_secret
-DISCORD_CALLBACK_URL=/oauth/discord/callback # this should be the same for everyone
+FIREBASE_API_KEY=
+FIREBASE_AUTH_DOMAIN=
+FIREBASE_PROJECT_ID=
+FIREBASE_STORAGE_BUCKET=
+FIREBASE_MESSAGING_SENDER_ID=
+FIREBASE_APP_ID=

-###########################
-# Application Domains
-###########################
+#========================#
+# Shared Links           #
+#========================#

-# Note: 
-# Server = Backend 
-# Client = Public (the client is the url you visit)
-# For the Google login to work in dev mode, you will need to change DOMAIN_SERVER to localhost:3090 or place it in .env.development
+ALLOW_SHARED_LINKS=true
+ALLOW_SHARED_LINKS_PUBLIC=true

-DOMAIN_CLIENT=http://localhost:3080
-DOMAIN_SERVER=http://localhost:3080
+#==============================#
+# Static File Cache Control    #
+#==============================#

-###########################
-# Email
-###########################
+# Leave commented out to use defaults: 1 day (86400 seconds) for s-maxage and 2 days (172800 seconds) for max-age
+# NODE_ENV must be set to production for these to take effect
+# STATIC_CACHE_MAX_AGE=172800
+# STATIC_CACHE_S_MAX_AGE=86400

-# Email is used for password reset. Note that all 4 values must be set for email to work.
-EMAIL_SERVICE=   # eg. gmail
-EMAIL_USERNAME=  # eg. your email address if using gmail
-EMAIL_PASSWORD=  # eg. this is the "app password" if using gmail
-EMAIL_FROM=      # eg. email address for from field like noreply@librechat.ai
+# If you have another service in front of your LibreChat doing compression, disable express based compression here
+# DISABLE_COMPRESSION=true
+
+#===================================================#
+#                        UI                         #
+#===================================================#
+
+APP_TITLE=LibreChat
+# CUSTOM_FOOTER="My custom footer"
+HELP_AND_FAQ_URL=https://librechat.ai
+
+# SHOW_BIRTHDAY_ICON=true
+
+# Google tag manager id
+#ANALYTICS_GTM_ID=user provided google tag manager id
+
+#==================================================#
+#                      Others                      #
+#==================================================#
+#   You should leave the following commented out   #
+
+# NODE_ENV=
+
+# REDIS_URI=
+# USE_REDIS=
+
+# E2E_USER_EMAIL=
+# E2E_USER_PASSWORD=
--- a/.eslintrc.js
+++ b/.eslintrc.js
@@ -12,6 +12,7 @@ module.exports = {
    'plugin:react-hooks/recommended',
    'plugin:jest/recommended',
    'prettier',
+    'plugin:jsx-a11y/recommended',
  ],
  ignorePatterns: [
    'client/dist/**/*',
@@ -19,6 +20,10 @@ module.exports = {
    'e2e/playwright-report/**/*',
    'packages/data-provider/types/**/*',
    'packages/data-provider/dist/**/*',
+    'packages/data-provider/test_bundle/**/*',
+    'data-node/**/*',
+    'meili_data/**/*',
+    'node_modules/**/*',
  ],
  parser: '@typescript-eslint/parser',
  parserOptions: {
@@ -28,7 +33,7 @@ module.exports = {
      jsx: true,
    },
  },
-  plugins: ['react', 'react-hooks', '@typescript-eslint', 'import'],
+  plugins: ['react', 'react-hooks', '@typescript-eslint', 'import', 'jsx-a11y'],
  rules: {
    'react/react-in-jsx-scope': 'off',
    '@typescript-eslint/ban-ts-comment': ['error', { 'ts-ignore': 'allow' }],
@@ -61,6 +66,8 @@ module.exports = {
    'no-restricted-syntax': 'off',
    'react/prop-types': ['off'],
    'react/display-name': ['off'],
+    'no-nested-ternary': 'error',
+    'no-unused-vars': ['error', { varsIgnorePattern: '^_' }],
    quotes: ['error', 'single'],
  },
  overrides: [
@@ -113,6 +120,8 @@ module.exports = {
      ],
      rules: {
        '@typescript-eslint/no-explicit-any': 'error',
+        '@typescript-eslint/no-unnecessary-condition': 'warn',
+        '@typescript-eslint/strict-boolean-expressions': 'warn',
      },
    },
    {
@@ -127,6 +136,19 @@ module.exports = {
        },
      ],
    },
+    {
+      files: './config/translations/**/*.ts',
+      parser: '@typescript-eslint/parser',
+      parserOptions: {
+        project: './config/translations/tsconfig.json',
+      },
+    },
+    {
+      files: ['./packages/data-provider/specs/**/*.ts'],
+      parserOptions: {
+        project: './packages/data-provider/tsconfig.spec.json',
+      },
+    },
  ],
  settings: {
    react: {
--- a/.github/CODE_OF_CONDUCT.md
+++ b/.github/CODE_OF_CONDUCT.md
@@ -60,7 +60,7 @@ representative at an online or offline event.

 Instances of abusive, harassing, or otherwise unacceptable behavior may be
 reported to the community leaders responsible for enforcement here on GitHub or
-on the official [Discord Server](https://discord.gg/uDyZ5Tzhct). 
+on the official [Discord Server](https://discord.librechat.ai). 
 All complaints will be reviewed and investigated promptly and fairly.

 All community leaders are obligated to respect the privacy and security of the
--- a/.github/CONTRIBUTING.md
+++ b/.github/CONTRIBUTING.md
@@ -8,7 +8,7 @@ If the feature you would like to contribute has not already received prior appro

 Please note that a pull request involving a feature that has not been reviewed and approved by the project maintainers may be rejected. We appreciate your understanding and cooperation.

-If you would like to discuss the changes you wish to make, join our [Discord community](https://discord.gg/uDyZ5Tzhct), where you can engage with other contributors and seek guidance from the community.
+If you would like to discuss the changes you wish to make, join our [Discord community](https://discord.librechat.ai), where you can engage with other contributors and seek guidance from the community.

 ## Our Standards

@@ -33,7 +33,7 @@ Project maintainers have the right and responsibility to remove, edit, or reject
 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: `tsc --noEmit`.
+    - 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`
@@ -126,6 +126,18 @@ 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
+
+- `npm` packages first, 
+     - from shortest line (top) to longest (bottom)
+
+- Followed by typescript types (pertains to data-provider and client workspaces)
+     - longest line (top) to shortest (bottom)
+     - types from package come first
+
+- Lastly, local imports
+     - longest line (top) to shortest (bottom)
+     - imports with alias `~` treated the same as relative import with respect to line length

 ---

--- a/.github/ISSUE_TEMPLATE/BUG-REPORT.yml
+++ b/.github/ISSUE_TEMPLATE/BUG-REPORT.yml
@@ -7,14 +7,6 @@ body:
    attributes:
      value: |
        Thanks for taking the time to fill out this bug report!
-  - type: input
-    id: contact
-    attributes:
-      label: Contact Details
-      description: How can we get in touch with you if we need more info?
-      placeholder: ex. email@example.com
-    validations:
-      required: false
  - type: textarea
    id: what-happened
    attributes:
@@ -58,7 +50,7 @@ body:
    id: terms
    attributes:
      label: Code of Conduct
-      description: By submitting this issue, you agree to follow our [Code of Conduct](https://github.com/danny-avila/LibreChat/blob/main/CODE_OF_CONDUCT.md)
+      description: By submitting this issue, you agree to follow our [Code of Conduct](https://github.com/danny-avila/LibreChat/blob/main/.github/CODE_OF_CONDUCT.md)
      options:
        - label: I agree to follow this project's Code of Conduct
          required: true
--- a/.github/ISSUE_TEMPLATE/FEATURE-REQUEST.yml
+++ b/.github/ISSUE_TEMPLATE/FEATURE-REQUEST.yml
@@ -7,14 +7,6 @@ body:
    attributes:
      value: |
        Thank you for taking the time to fill this out!
-  - type: input
-    id: contact
-    attributes:
-      label: Contact Details
-      description: How can we contact you if we need more information?
-      placeholder: ex. email@example.com
-    validations:
-      required: false
  - type: textarea
    id: what
    attributes:
@@ -51,7 +43,7 @@ body:
    id: terms
    attributes:
      label: Code of Conduct
-      description: By submitting this issue, you agree to follow our [Code of Conduct](https://github.com/danny-avila/LibreChat/blob/main/CODE_OF_CONDUCT.md)
+      description: By submitting this issue, you agree to follow our [Code of Conduct](https://github.com/danny-avila/LibreChat/blob/main/.github/CODE_OF_CONDUCT.md)
      options:
        - label: I agree to follow this project's Code of Conduct
          required: true
--- a/.github/ISSUE_TEMPLATE/QUESTION.yml
+++ b/.github/ISSUE_TEMPLATE/QUESTION.yml
@@ -7,14 +7,6 @@ body:
    attributes:
      value: |
        Thanks for taking the time to fill this!
-  - type: input
-    id: contact
-    attributes:
-      label: Contact Details
-      description: How can we get in touch with you if we need more info?
-      placeholder: ex. email@example.com
-    validations:
-      required: false
  - type: textarea
    id: what-is-your-question
    attributes:
@@ -52,7 +44,7 @@ body:
    id: terms
    attributes:
      label: Code of Conduct
-      description: By submitting this issue, you agree to follow our [Code of Conduct](https://github.com/danny-avila/LibreChat/blob/main/CODE_OF_CONDUCT.md)
+      description: By submitting this issue, you agree to follow our [Code of Conduct](https://github.com/danny-avila/LibreChat/blob/main/.github/CODE_OF_CONDUCT.md)
      options:
        - label: I agree to follow this project's Code of Conduct
          required: true
--- a/.github/SECURITY.md
+++ b/.github/SECURITY.md
@@ -12,7 +12,7 @@ When reporting a security vulnerability, you have the following options to reach

 - **Option 2: GitHub Issues**: You can initiate first contact via GitHub Issues. However, please note that initial contact through GitHub Issues should not include any sensitive details.

- **Option 3: Discord Server**: You can join our [Discord community](https://discord.gg/5rbRxn4uME) and initiate first contact in the `#issues` channel. However, please ensure that initial contact through Discord does not include any sensitive details.
+- **Option 3: Discord Server**: You can join our [Discord community](https://discord.librechat.ai) and initiate first contact in the `#issues` channel. However, please ensure that initial contact through Discord does not include any sensitive details.

 _After the initial contact, we will establish a private communication channel for further discussion._

@@ -39,11 +39,11 @@ Please note that as a security-conscious community, we may not always disclose d

 This security policy applies to the following GitHub repository:

- Repository: [LibreChat](https://github.com/danny-avila/LibreChat)
+- Repository: [LibreChat](https://github.librechat.ai)

 ## Contact

-If you have any questions or concerns regarding the security of our project, please join our [Discord community](https://discord.gg/NGaa9RPCft) and report them in the appropriate channel. You can also reach out to us by [opening an issue](https://github.com/danny-avila/LibreChat/issues/new) on GitHub. Please note that the response time may vary depending on the nature and severity of the inquiry.
+If you have any questions or concerns regarding the security of our project, please join our [Discord community](https://discord.librechat.ai) and report them in the appropriate channel. You can also reach out to us by [opening an issue](https://github.com/danny-avila/LibreChat/issues/new) on GitHub. Please note that the response time may vary depending on the nature and severity of the inquiry.

 ## Acknowledgments

--- a/.github/playwright.yml
+++ b/.github/playwright.yml
@@ -0,0 +1,72 @@
+# name: Playwright Tests
+# on:
+#   pull_request:
+#     branches: 
+#       - main
+#       - dev
+#       - release/*
+#     paths:
+#       - 'api/**'
+#       - 'client/**'
+#       - 'packages/**'
+#       - 'e2e/**'
+# jobs:
+#   tests_e2e:
+#     name: Run Playwright tests
+#     if: github.event.pull_request.head.repo.full_name == 'danny-avila/LibreChat'
+#     timeout-minutes: 60
+#     runs-on: ubuntu-latest
+#     env:
+#       NODE_ENV: CI
+#       CI: true
+#       SEARCH: false
+#       BINGAI_TOKEN: user_provided
+#       CHATGPT_TOKEN: user_provided
+#       MONGO_URI: ${{ secrets.MONGO_URI }}
+#       OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
+#       E2E_USER_EMAIL: ${{ secrets.E2E_USER_EMAIL }}
+#       E2E_USER_PASSWORD: ${{ secrets.E2E_USER_PASSWORD }}
+#       JWT_SECRET: ${{ secrets.JWT_SECRET }}
+#       JWT_REFRESH_SECRET: ${{ secrets.JWT_REFRESH_SECRET }}
+#       CREDS_KEY: ${{ secrets.CREDS_KEY }}
+#       CREDS_IV: ${{ secrets.CREDS_IV }}
+#       DOMAIN_CLIENT: ${{ secrets.DOMAIN_CLIENT }}
+#       DOMAIN_SERVER: ${{ secrets.DOMAIN_SERVER }}
+#       PLAYWRIGHT_SKIP_BROWSER_DOWNLOAD: 1 # Skip downloading during npm install
+#       PLAYWRIGHT_BROWSERS_PATH: 0 # Places binaries to node_modules/@playwright/test
+#       TITLE_CONVO: false
+#     steps:
+#       - uses: actions/checkout@v4
+#       - uses: actions/setup-node@v4
+#         with:
+#           node-version: 18
+#           cache: 'npm'
+
+#       - name: Install global dependencies
+#         run: npm ci
+
+#       # - name: Remove sharp dependency
+#       #   run: rm -rf node_modules/sharp
+
+#       # - name: Install sharp with linux dependencies
+#       #   run: cd api && SHARP_IGNORE_GLOBAL_LIBVIPS=1 npm install --arch=x64 --platform=linux --libc=glibc sharp
+
+#       - name: Build Client
+#         run: npm run frontend
+
+#       - name: Install Playwright
+#         run: |
+#           npx playwright install-deps
+#           npm install -D @playwright/test@latest
+#           npx playwright install chromium
+
+#       - name: Run Playwright tests
+#         run: npm run e2e:ci
+
+#       - name: Upload playwright report
+#         uses: actions/upload-artifact@v3
+#         if: always()
+#         with:
+#           name: playwright-report
+#           path: e2e/playwright-report/
+#           retention-days: 30
--- a/.github/pull_request_template.md
+++ b/.github/pull_request_template.md
@@ -1,7 +1,10 @@
 # Pull Request Template

+⚠️ Before Submitting a PR, Please Review:
+- Please ensure that you have thoroughly read and understood the [Contributing Docs](https://github.com/danny-avila/LibreChat/blob/main/.github/CONTRIBUTING.md) before submitting your Pull Request.

-### ⚠️ Before Submitting a PR, read the [Contributing Docs](./CONTRIBUTING.md) in full!
+⚠️ Documentation Updates Notice:
+- Kindly note that documentation updates are managed in this repository: [librechat.ai](https://github.com/LibreChat-AI/librechat.ai)

 ## Summary

@@ -15,7 +18,7 @@ Please delete any irrelevant options.
 - [ ] New feature (non-breaking change which adds functionality)
 - [ ] Breaking change (fix or feature that would cause existing functionality to not work as expected)
 - [ ] This change requires a documentation update
- [ ] Documentation update 
+- [ ] Translation update

 ## Testing

@@ -25,6 +28,8 @@ Please describe your test process and include instructions so that we can reprod

 ## Checklist

+Please delete any irrelevant options.
+
 - [ ] My code adheres to this project's style guidelines
 - [ ] I have performed a self-review of my own code
 - [ ] I have commented in any complex areas of my code
@@ -33,3 +38,4 @@ Please describe your test process and include instructions so that we can reprod
 - [ ] I have written tests demonstrating that my changes are effective or that my feature works
 - [ ] Local unit tests pass with my changes
 - [ ] Any changes dependent on mine have been merged and published in downstream modules.
+- [ ] A pull request for updating the documentation has been submitted.
--- a/.github/workflows/a11y.yml
+++ b/.github/workflows/a11y.yml
@@ -0,0 +1,26 @@
+name: Lint for accessibility issues
+
+on:
+  pull_request:
+    paths:
+      - 'client/src/**'
+  workflow_dispatch:
+    inputs:
+      run_workflow:
+        description: 'Set to true to run this workflow'
+        required: true
+        default: 'false'
+
+jobs:
+  axe-linter:
+    runs-on: ubuntu-latest
+    if: >
+      (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == 'danny-avila/LibreChat') ||
+      (github.event_name == 'workflow_dispatch' && github.event.inputs.run_workflow == 'true')
+    
+    steps:
+      - uses: actions/checkout@v4
+      - uses: dequelabs/axe-linter-action@v1
+        with:
+          api_key: ${{ secrets.AXE_LINTER_API_KEY }}
+          github_token: ${{ secrets.GITHUB_TOKEN }}
--- a/.github/workflows/backend-review.yml
+++ b/.github/workflows/backend-review.yml
@@ -21,11 +21,11 @@ jobs:
      BAN_VIOLATIONS: ${{ secrets.BAN_VIOLATIONS }}
      BAN_DURATION: ${{ secrets.BAN_DURATION }}
      BAN_INTERVAL: ${{ secrets.BAN_INTERVAL }}
-      NODE_ENV: ci
+      NODE_ENV: CI
    steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v4
      - name: Use Node.js 20.x
-        uses: actions/setup-node@v3
+        uses: actions/setup-node@v4
        with:
          node-version: 20
          cache: 'npm'
@@ -35,11 +35,32 @@ jobs:

      - name: Install Data Provider
        run: npm run build:data-provider
+      
+      - name: Create empty auth.json file
+        run: |
+          mkdir -p api/data
+          echo '{}' > api/data/auth.json
+
+      - name: Check for Circular dependency in rollup
+        working-directory: ./packages/data-provider
+        run: |
+          output=$(npm run rollup:api)
+          echo "$output"
+          if echo "$output" | grep -q "Circular dependency"; then
+            echo "Error: Circular dependency detected!"
+            exit 1
+          fi
+
+      - name: Prepare .env.test file
+        run: cp api/test/.env.test.example api/test/.env.test

      - name: Run unit tests
        run: cd api && npm run test:ci

+      - name: Run librechat-data-provider unit tests
+        run: cd packages/data-provider && npm run test:ci
+
      - name: Run linters
        uses: wearerequired/lint-action@v2
        with:
-          eslint: true
+          eslint: true
--- a/.github/workflows/container.yml
+++ b/.github/workflows/container.yml
@@ -1,52 +0,0 @@
-name: Docker Compose Build on Tag
-
-# The workflow is triggered when a tag is pushed
-on:
-  push:
-    tags:
-      - "*"
-
-jobs:
-  build:
-    runs-on: ubuntu-latest
-
-    steps:
-      # Check out the repository
-      - name: Checkout
-        uses: actions/checkout@v2
-
-      # Set up Docker
-      - name: Set up Docker
-        uses: docker/setup-buildx-action@v1
-
-      # 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: ${{ secrets.GITHUB_TOKEN }}
-
-      # Run docker-compose build
-      - name: Build Docker images
-        run: |
-          cp .env.example .env
-          docker-compose build
-          docker build -f Dockerfile.multi --target api-build -t librechat-api .
-
-      # Get Tag Name
-      - name: Get Tag Name
-        id: tag_name
-        run: echo "TAG_NAME=${GITHUB_REF/refs\/tags\//}" >> $GITHUB_ENV
-
-      # Tag it properly before push to github
-      - name: tag image and push
-        run: |
-          docker tag librechat:latest ghcr.io/${{ github.repository_owner }}/librechat:${{ env.TAG_NAME }}
-          docker push ghcr.io/${{ github.repository_owner }}/librechat:${{ env.TAG_NAME }}
-          docker tag librechat:latest ghcr.io/${{ github.repository_owner }}/librechat:latest
-          docker push ghcr.io/${{ github.repository_owner }}/librechat:latest
-          docker tag librechat-api:latest ghcr.io/${{ github.repository_owner }}/librechat-api:${{ env.TAG_NAME }}
-          docker push ghcr.io/${{ github.repository_owner }}/librechat-api:${{ env.TAG_NAME }}
-          docker tag librechat-api:latest ghcr.io/${{ github.repository_owner }}/librechat-api:latest
-          docker push ghcr.io/${{ github.repository_owner }}/librechat-api:latest
--- a/.github/workflows/data-provider.yml
+++ b/.github/workflows/data-provider.yml
@@ -11,8 +11,8 @@ jobs:
  build:
    runs-on: ubuntu-latest
    steps:
-      - uses: actions/checkout@v3
-      - uses: actions/setup-node@v3
+      - uses: actions/checkout@v4
+      - uses: actions/setup-node@v4
        with:
          node-version: 16
      - run: cd packages/data-provider && npm ci
@@ -22,8 +22,8 @@ jobs:
    needs: build
    runs-on: ubuntu-latest
    steps:
-      - uses: actions/checkout@v3
-      - uses: actions/setup-node@v3
+      - uses: actions/checkout@v4
+      - uses: actions/setup-node@v4
        with:
          node-version: 16
          registry-url: 'https://registry.npmjs.org'
--- a/.github/workflows/deploy-dev.yml
+++ b/.github/workflows/deploy-dev.yml
@@ -0,0 +1,41 @@
+name: Update Test Server
+
+on:
+  workflow_run:
+    workflows: ["Docker Dev Images Build"]
+    types:
+      - completed
+  workflow_dispatch:
+
+jobs:
+  deploy:
+    runs-on: ubuntu-latest
+    if: |
+      github.repository == 'danny-avila/LibreChat' &&
+      (github.event_name == 'workflow_dispatch' || github.event.workflow_run.conclusion == 'success')
+    steps:
+    - name: Checkout repository
+      uses: actions/checkout@v4
+
+    - name: Install SSH Key
+      uses: shimataro/ssh-key-action@v2
+      with:
+        key: ${{ secrets.DO_SSH_PRIVATE_KEY }}
+        known_hosts: ${{ secrets.DO_KNOWN_HOSTS }}
+
+    - name: Run update script on DigitalOcean Droplet
+      env:
+        DO_HOST: ${{ secrets.DO_HOST }}
+        DO_USER: ${{ secrets.DO_USER }}
+      run: |
+        ssh -o StrictHostKeyChecking=no ${DO_USER}@${DO_HOST} << EOF
+        sudo -i -u danny bash << EEOF
+        cd ~/LibreChat && \
+        git fetch origin main && \
+        npm run update:deployed && \
+        git checkout do-deploy && \
+        git rebase main && \
+        npm run start:deployed && \
+        echo "Update completed. Application should be running now."
+        EEOF
+        EOF
--- a/.github/workflows/deploy.yml
+++ b/.github/workflows/deploy.yml
@@ -17,7 +17,7 @@ jobs:
    steps:
      # checkout the repo
      - name: 'Checkout GitHub Action'
-        uses: actions/checkout@main
+        uses: actions/checkout@v4

      - name: 'Login via Azure CLI'
        uses: azure/login@v1
--- a/.github/workflows/dev-images.yml
+++ b/.github/workflows/dev-images.yml
@@ -8,19 +8,33 @@ on:
    paths:
      - 'api/**'
      - 'client/**'
+      - 'packages/**'

 jobs:
  build:
    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        include:
+          - target: api-build
+            file: Dockerfile.multi
+            image_name: librechat-dev-api
+          - target: node
+            file: Dockerfile
+            image_name: librechat-dev

    steps:
      # Check out the repository
      - name: Checkout
-        uses: actions/checkout@v2
+        uses: actions/checkout@v4

-      # Set up Docker
-      - name: Set up Docker
-        uses: docker/setup-buildx-action@v1
+      # 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
@@ -30,22 +44,29 @@ jobs:
          username: ${{ github.actor }}
          password: ${{ secrets.GITHUB_TOKEN }}

-      # Build Docker images
-      - name: Build Docker images
+      # 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
-          docker build -f Dockerfile.multi --target api-build -t librechat-dev-api .
-          docker build -f Dockerfile -t librechat-dev .

-      # Tag and push the images to GitHub Container Registry
-      - name: Tag and push images
-        run: |
-          docker tag librechat-dev-api:latest ghcr.io/${{ github.repository_owner }}/librechat-dev-api:${{ github.sha }}
-          docker push ghcr.io/${{ github.repository_owner }}/librechat-dev-api:${{ github.sha }}
-          docker tag librechat-dev-api:latest ghcr.io/${{ github.repository_owner }}/librechat-dev-api:latest
-          docker push ghcr.io/${{ github.repository_owner }}/librechat-dev-api:latest
-
-          docker tag librechat-dev:latest ghcr.io/${{ github.repository_owner }}/librechat-dev:${{ github.sha }}
-          docker push ghcr.io/${{ github.repository_owner }}/librechat-dev:${{ github.sha }}
-          docker tag librechat-dev:latest ghcr.io/${{ github.repository_owner }}/librechat-dev:latest
-          docker push ghcr.io/${{ github.repository_owner }}/librechat-dev:latest
+      # 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
@@ -1,11 +1,6 @@
-#github action to run unit tests for frontend with jest
 name: Frontend Unit Tests
+
 on:
-  # push:
-  #   branches: 
-  #     - main
-  #     - dev
-  #     - release/*
  pull_request:
    branches: 
      - main
@@ -14,15 +9,16 @@ on:
    paths:
      - 'client/**'
      - 'packages/**'
+
 jobs:
-  tests_frontend:
-    name: Run frontend unit tests
+  tests_frontend_ubuntu:
+    name: Run frontend unit tests on Ubuntu
    timeout-minutes: 60
    runs-on: ubuntu-latest
    steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v4
      - name: Use Node.js 20.x
-        uses: actions/setup-node@v3
+        uses: actions/setup-node@v4
        with:
          node-version: 20
          cache: 'npm'
@@ -34,4 +30,27 @@ jobs:
        run: npm run frontend:ci

      - name: Run unit tests
-        run: cd client && npm run test:ci
+        run: npm run test:ci --verbose
+        working-directory: client
+
+  tests_frontend_windows:
+    name: Run frontend unit tests on Windows
+    timeout-minutes: 60
+    runs-on: windows-latest
+    steps:
+      - uses: actions/checkout@v4
+      - name: Use Node.js 20.x
+        uses: actions/setup-node@v4
+        with:
+          node-version: 20
+          cache: 'npm'
+
+      - name: Install dependencies
+        run: npm ci
+
+      - name: Build Client
+        run: npm run frontend:ci
+
+      - name: Run unit tests
+        run: npm run test:ci --verbose
+        working-directory: client
--- a/.github/workflows/generate_embeddings.yml
+++ b/.github/workflows/generate_embeddings.yml
@@ -0,0 +1,20 @@
+name: 'generate_embeddings'
+on:
+  workflow_dispatch:
+  push:
+    branches:
+      - main
+    paths:
+      - 'docs/**'
+
+jobs:
+  generate:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v3
+      - uses: supabase/embeddings-generator@v0.0.5
+        with:
+          supabase-url: ${{ secrets.SUPABASE_URL }}
+          supabase-service-role-key: ${{ secrets.SUPABASE_SERVICE_ROLE_KEY }}
+          openai-key: ${{ secrets.OPENAI_DOC_EMBEDDINGS_KEY }}
+          docs-root-path: 'docs'
--- a/.github/workflows/helmcharts.yml
+++ b/.github/workflows/helmcharts.yml
@@ -0,0 +1,35 @@
+name: Build Helm Charts on Tag
+
+# The workflow is triggered when a tag is pushed
+on:
+  push:
+    tags:
+      - "*"
+
+jobs:
+  release:
+    permissions:
+      contents: write
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Configure Git
+        run: |
+          git config user.name "$GITHUB_ACTOR"
+          git config user.email "$GITHUB_ACTOR@users.noreply.github.com"
+
+      - name: Install Helm
+        uses: azure/setup-helm@v4
+        env:
+          GITHUB_TOKEN: "${{ secrets.GITHUB_TOKEN }}"       
+
+      - name: Run chart-releaser
+        uses: helm/chart-releaser-action@v1.6.0
+        with:
+          charts_dir: helmchart
+        env:
+          CR_TOKEN: "${{ secrets.GITHUB_TOKEN }}"
--- a/.github/workflows/main-image-workflow.yml
+++ b/.github/workflows/main-image-workflow.yml
@@ -0,0 +1,69 @@
+name: Docker Compose Build Latest Main Image Tag (Manual Dispatch)
+
+on:
+  workflow_dispatch:
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        include:
+          - target: api-build
+            file: Dockerfile.multi
+            image_name: librechat-api
+          - target: node
+            file: Dockerfile
+            image_name: librechat
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Fetch tags and set the latest tag
+        run: |
+          git fetch --tags
+          echo "LATEST_TAG=$(git describe --tags `git rev-list --tags --max-count=1`)" >> $GITHUB_ENV
+
+      # 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 }}/${{ matrix.image_name }}:${{ env.LATEST_TAG }}
+            ghcr.io/${{ github.repository_owner }}/${{ matrix.image_name }}:latest
+            ${{ secrets.DOCKERHUB_USERNAME }}/${{ matrix.image_name }}:${{ env.LATEST_TAG }}
+            ${{ secrets.DOCKERHUB_USERNAME }}/${{ matrix.image_name }}:latest
+          platforms: linux/amd64,linux/arm64
+          target: ${{ matrix.target }}
--- a/.github/workflows/mkdocs.yaml
+++ b/.github/workflows/mkdocs.yaml
@@ -1,24 +0,0 @@
-name: mkdocs
-on:
-  push:
-    branches:
-      - main
-permissions:
-  contents: write
-jobs:
-  deploy:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v3
-      - uses: actions/setup-python@v4
-        with:
-          python-version: 3.x
-      - run: echo "cache_id=$(date --utc '+%V')" >> $GITHUB_ENV 
-      - uses: actions/cache@v3
-        with:
-          key: mkdocs-material-${{ env.cache_id }}
-          path: .cache
-          restore-keys: |
-            mkdocs-material-
-      - run: pip install mkdocs-material 
-      - run: mkdocs gh-deploy --force
--- a/.github/workflows/playwright.yml
+++ b/.github/workflows/playwright.yml
@@ -1,71 +0,0 @@
-name: Playwright Tests
-on:
-  pull_request:
-    branches: 
-      - main
-      - dev
-      - release/*
-    paths:
-      - 'api/**'
-      - 'client/**'
-      - 'packages/**'
-      - 'e2e/**'
-jobs:
-  tests_e2e:
-    name: Run Playwright tests
-    if: github.event.pull_request.head.repo.full_name == 'danny-avila/LibreChat'
-    timeout-minutes: 60
-    runs-on: ubuntu-latest
-    env:
-      NODE_ENV: ci
-      CI: true
-      SEARCH: false
-      BINGAI_TOKEN: user_provided
-      CHATGPT_TOKEN: user_provided
-      MONGO_URI: ${{ secrets.MONGO_URI }}
-      OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
-      E2E_USER_EMAIL: ${{ secrets.E2E_USER_EMAIL }}
-      E2E_USER_PASSWORD: ${{ secrets.E2E_USER_PASSWORD }}
-      JWT_SECRET: ${{ secrets.JWT_SECRET }}
-      JWT_REFRESH_SECRET: ${{ secrets.JWT_REFRESH_SECRET }}
-      CREDS_KEY: ${{ secrets.CREDS_KEY }}
-      CREDS_IV: ${{ secrets.CREDS_IV }}
-      DOMAIN_CLIENT: ${{ secrets.DOMAIN_CLIENT }}
-      DOMAIN_SERVER: ${{ secrets.DOMAIN_SERVER }}
-      PLAYWRIGHT_SKIP_BROWSER_DOWNLOAD: 1 # Skip downloading during npm install
-      PLAYWRIGHT_BROWSERS_PATH: 0 # Places binaries to node_modules/@playwright/test
-    steps:
-      - uses: actions/checkout@v3
-      - uses: actions/setup-node@v3
-        with:
-          node-version: 18
-          cache: 'npm'
-
-      - name: Install global dependencies
-        run: npm ci
-
-      # - name: Remove sharp dependency
-      #   run: rm -rf node_modules/sharp
-
-      # - name: Install sharp with linux dependencies
-      #   run: cd api && SHARP_IGNORE_GLOBAL_LIBVIPS=1 npm install --arch=x64 --platform=linux --libc=glibc sharp
-
-      - name: Build Client
-        run: npm run frontend
-
-      - name: Install Playwright
-        run: |
-          npx playwright install-deps
-          npm install -D @playwright/test@latest
-          npx playwright install chromium
-
-      - name: Run Playwright tests
-        run: npm run e2e:ci
-
-      - name: Upload playwright report
-        uses: actions/upload-artifact@v3
-        if: always()
-        with:
-          name: playwright-report
-          path: e2e/playwright-report/
-          retention-days: 30
--- a/.github/workflows/tag-images.yml
+++ b/.github/workflows/tag-images.yml
@@ -0,0 +1,67 @@
+name: Docker Images Build on Tag
+
+on:
+  push:
+    tags:
+      - '*'
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        include:
+          - target: api-build
+            file: Dockerfile.multi
+            image_name: librechat-api
+          - target: node
+            file: Dockerfile
+            image_name: librechat
+
+    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.ref_name }}
+            ghcr.io/${{ github.repository_owner }}/${{ matrix.image_name }}:latest
+            ${{ secrets.DOCKERHUB_USERNAME }}/${{ matrix.image_name }}:${{ github.ref_name }}
+            ${{ secrets.DOCKERHUB_USERNAME }}/${{ matrix.image_name }}:latest
+          platforms: linux/amd64,linux/arm64
+          target: ${{ matrix.target }}
--- a/.gitignore
+++ b/.gitignore
@@ -2,7 +2,7 @@

 # Logs
 data-node
-meili_data
+meili_data*
 data/
 logs
 *.log
@@ -11,6 +11,7 @@ logs
 pids
 *.pid
 *.seed
+.git

 # Directory for instrumented libs generated by jscoverage/JSCover
 lib-cov
@@ -21,6 +22,10 @@ coverage
 # Grunt intermediate storage (http://gruntjs.com/creating-plugins#storing-task-files)
 .grunt

+# translation services
+config/translations/stores/*
+client/src/localization/languages/*_missing_keys.json
+
 # Compiled Dirs (http://nodejs.org/api/addons.html)
 build/
 dist/
@@ -40,7 +45,8 @@ meili_data/
 api/node_modules/
 client/node_modules/
 bower_components/
-types/
+*.d.ts
+!vite-env.d.ts

 # Floobits
 .floo
@@ -48,6 +54,10 @@ types/
 .floo
 .flooignore

+#config file
+librechat.yaml
+librechat.yml
+
 # Environment
 .npmrc
 .env*
@@ -65,11 +75,20 @@ src/style - official.css
 /playwright/.cache/
 .DS_Store
 *.code-workspace
+.idx
+monospace.json
 .idea
+*.iml
 *.pem
 config.local.ts
 **/storageState.json
 junit.xml
+**/.venv/
+**/venv/
+
+# docker override file
+docker-compose.override.yaml
+docker-compose.override.yml

 # meilisearch
 meilisearch
@@ -78,4 +97,12 @@ data.ms/*
 auth.json

 /packages/ux-shared/
-/images
+/images
+
+!client/src/components/Nav/SettingsTabs/Data/
+
+# User uploads
+uploads/
+
+# owner
+release/
--- a/.husky/pre-commit
+++ b/.husky/pre-commit
@@ -1,4 +1,4 @@
-#!/usr/bin/env sh
+#!/usr/bin/env sh 
 set -e
 . "$(dirname -- "$0")/_/husky.sh"
 [ -n "$CI" ] && exit 0
--- a/.vscode/launch.json
+++ b/.vscode/launch.json
@@ -0,0 +1,16 @@
+{
+  "version": "0.2.0",
+  "configurations": [
+    {
+      "type": "node",
+      "request": "launch",
+      "name": "Launch LibreChat (debug)",
+      "skipFiles": ["<node_internals>/**"],
+      "program": "${workspaceFolder}/api/server/index.js",
+      "env": {
+        "NODE_ENV": "production"
+      },
+      "console": "integratedTerminal"
+    }
+  ]
+}
--- a/39
+++ b/39
@@ -1,21 +1,32 @@
-# Base node image
-FROM node:19-alpine AS node
+# v0.7.5-rc1

-COPY . /app
+# Base node image
+FROM node:20-alpine AS node
+
+RUN apk --no-cache add curl
+
+RUN mkdir -p /app && chown node:node /app
 WORKDIR /app

-# Install call deps - Install curl for health check
-RUN apk --no-cache add curl && \
-    # We want to inherit env from the container, not the file
-    # This will preserve any existing env file if it's already in souce
-    # otherwise it will create a new one
-    touch .env && \
-    # Build deps in seperate 
-    npm ci
+USER node

-# React client build
-ENV NODE_OPTIONS="--max-old-space-size=2048"
-RUN npm run frontend
+COPY --chown=node:node . .
+
+RUN \
+    # Allow mounting of these files, which have no default
+    touch .env ; \
+    # Create directories for the volumes to inherit the correct permissions
+    mkdir -p /app/client/public/images /app/api/logs ; \
+    npm config set fetch-retry-maxtimeout 600000 ; \
+    npm config set fetch-retries 5 ; \
+    npm config set fetch-retry-mintimeout 15000 ; \
+    npm install --no-audit; \
+    # React client build
+    NODE_OPTIONS="--max-old-space-size=2048" npm run frontend; \
+    npm prune --production; \
+    npm cache clean --force
+
+RUN mkdir -p /app/client/public/images /app/api/logs

 # Node API setup
 EXPOSE 3080
--- a/Dockerfile.multi
+++ b/Dockerfile.multi
@@ -1,40 +1,44 @@
-# Build API, Client and Data Provider
-FROM node:19-alpine AS base
+# Dockerfile.multi
+# v0.7.5-rc1

+# Base for all builds
+FROM node:20-alpine AS base
 WORKDIR /app
-COPY config/loader.js ./config/
-RUN npm install dotenv
+RUN apk --no-cache add curl
+RUN npm config set fetch-retry-maxtimeout 600000 && \
+    npm config set fetch-retries 5 && \
+    npm config set fetch-retry-mintimeout 15000
+COPY package*.json ./
+COPY packages/data-provider/package*.json ./packages/data-provider/
+COPY client/package*.json ./client/
+COPY api/package*.json ./api/
+RUN npm ci

-WORKDIR /app/api
-COPY api/package*.json ./
-COPY api/ ./
-RUN npm install
+# Build data-provider
+FROM base AS data-provider-build
+WORKDIR /app/packages/data-provider
+COPY packages/data-provider ./
+RUN npm run build
+RUN npm prune --production

-# React client build
+# Client build
 FROM base AS client-build
 WORKDIR /app/client
-COPY ./client/ ./
-
-WORKDIR /app/packages/data-provider
-COPY ./packages/data-provider ./
-RUN npm install
-RUN npm run build
-RUN mkdir -p /app/client/node_modules/librechat-data-provider/
-RUN cp -R /app/packages/data-provider/* /app/client/node_modules/librechat-data-provider/
-
-WORKDIR /app/client
-RUN npm install
+COPY client ./
+COPY --from=data-provider-build /app/packages/data-provider/dist /app/packages/data-provider/dist
 ENV NODE_OPTIONS="--max-old-space-size=2048"
 RUN npm run build
+RUN npm prune --production

-# Node API setup
+# API setup (including client dist)
 FROM base AS api-build
-COPY --from=client-build /app/client/dist /app/client/dist
+WORKDIR /app
+COPY api ./api
+COPY config ./config
+COPY --from=data-provider-build /app/packages/data-provider/dist ./packages/data-provider/dist
+COPY --from=client-build /app/client/dist ./client/dist
+WORKDIR /app/api
+RUN npm prune --production
 EXPOSE 3080
 ENV HOST=0.0.0.0
 CMD ["node", "server/index.js"]
-
-# Nginx setup
-FROM nginx:1.21.1-alpine AS prod-stage
-COPY ./client/nginx.conf /etc/nginx/conf.d/default.conf
-CMD ["nginx", "-g", "daemon off;"]
--- a/.github/LICENSE.md
+++ b/.github/LICENSE.md
@@ -1,8 +1,6 @@
-# MIT License
+MIT License

-Copyright (c) 2023 Danny Avila
-
---
+Copyright (c) 2024 LibreChat

 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal
@@ -14,8 +12,6 @@ furnished to do so, subject to the following conditions:
 The above copyright notice and this permission notice shall be included in all
 copies or substantial portions of the Software.

-##
-
 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
@@ -23,7 +19,3 @@ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 SOFTWARE.
-
---
-
-## [Go Back to ReadMe](../README.md)
--- a/README.md
+++ b/README.md
@@ -1,14 +1,14 @@
 <p align="center">
-  <a href="https://docs.librechat.ai">
-    <img src="docs/assets/LibreChat.svg" height="256">
-  </a>
-  <a href="https://docs.librechat.ai">
-    <h1 align="center">LibreChat</h1>
+  <a href="https://librechat.ai">
+    <img src="client/public/assets/logo.svg" height="256">
  </a>
+  <h1 align="center">
+    <a href="https://librechat.ai">LibreChat</a>
+  </h1>
 </p>

 <p align="center">
-  <a href="https://discord.gg/NGaa9RPCft"> 
+  <a href="https://discord.librechat.ai"> 
    <img
      src="https://img.shields.io/discord/1086345563026489514?label=&logo=discord&style=for-the-badge&logoWidth=20&logoColor=white&labelColor=000000&color=blueviolet">
  </a>
@@ -20,141 +20,123 @@
    <img
      src="https://img.shields.io/badge/DOCS-blue.svg?style=for-the-badge&logo=read-the-docs&logoColor=white&labelColor=000000&logoWidth=20">
  </a>
-  <a aria-label="Sponsors" href="#sponsors">
+  <a aria-label="Sponsors" href="https://github.com/sponsors/danny-avila">
    <img
      src="https://img.shields.io/badge/SPONSORS-brightgreen.svg?style=for-the-badge&logo=github-sponsors&logoColor=white&labelColor=000000&logoWidth=20">
  </a>
 </p>

-## All-In-One AI Conversations with LibreChat ##
+<p align="center">
+<a href="https://railway.app/template/b5k2mn?referralCode=HI9hWz">
+  <img src="https://railway.app/button.svg" alt="Deploy on Railway" height="30">
+</a>
+<a href="https://zeabur.com/templates/0X2ZY8">
+  <img src="https://zeabur.com/button.svg" alt="Deploy on Zeabur" height="30"/>
+</a>
+<a href="https://template.cloud.sealos.io/deploy?templateName=librechat">
+  <img src="https://raw.githubusercontent.com/labring-actions/templates/main/Deploy-on-Sealos.svg" alt="Deploy on Sealos" height="30">
+</a>
+</p>
+
+# 📃 Features
+
+- 🖥️ UI matching ChatGPT, including Dark mode, Streaming, and latest updates
+- 🤖 AI model selection:
+  - OpenAI, Azure OpenAI, BingAI, ChatGPT, Google Vertex AI, Anthropic (Claude), Plugins, Assistants API (including Azure Assistants)
+- ✅ Compatible across both **[Remote & Local AI services](https://www.librechat.ai/docs/configuration/librechat_yaml/ai_endpoints):**
+  - groq, Ollama, Cohere, Mistral AI, Apple MLX, koboldcpp, OpenRouter, together.ai, Perplexity, ShuttleAI, and more
+- 💾 Create, Save, & Share Custom Presets
+- 🔀 Switch between AI Endpoints and Presets, mid-chat
+- 🔄 Edit, Resubmit, and Continue Messages with Conversation branching
+- 🌿 Fork Messages & Conversations for Advanced Context control
+- 💬 Multimodal Chat:
+    - Upload and analyze images with Claude 3, GPT-4 (including `gpt-4o` and `gpt-4o-mini`), and Gemini Vision 📸
+    - Chat with Files using Custom Endpoints, OpenAI, Azure, Anthropic, & Google. 🗃️
+    - Advanced Agents with Files, Code Interpreter, Tools, and API Actions 🔦
+      - Available through the [OpenAI Assistants API](https://platform.openai.com/docs/assistants/overview) 🌤️
+      - Non-OpenAI Agents in Active Development 🚧
+- 🌎 Multilingual UI:
+  - English, 中文, Deutsch, Español, Français, Italiano, Polski, Português Brasileiro,
+  - Русский, 日本語, Svenska, 한국어, Tiếng Việt, 繁體中文, العربية, Türkçe, Nederlands, עברית
+- 🎨 Customizable Dropdown & Interface: Adapts to both power users and newcomers
+- 📧 Verify your email to ensure secure access
+- 🗣️ Chat hands-free with Speech-to-Text and Text-to-Speech magic
+  - Automatically send and play Audio
+  - Supports OpenAI, Azure OpenAI, and Elevenlabs
+- 📥 Import Conversations from LibreChat, ChatGPT, Chatbot UI
+- 📤 Export conversations as screenshots, markdown, text, json
+- 🔍 Search all messages/conversations
+- 🔌 Plugins, including web access, image generation with DALL-E-3 and more
+- 👥 Multi-User, Secure Authentication with Moderation and Token spend tools
+- ⚙️ Configure Proxy, Reverse Proxy, Docker, & many Deployment options:
+  - Use completely local or deploy on the cloud
+- 📖 Completely Open-Source & Built in Public
+- 🧑‍🤝‍🧑 Community-driven development, support, and feedback
+
+[For a thorough review of our features, see our docs here](https://docs.librechat.ai/) 📚
+
+## 🪶 All-In-One AI Conversations with LibreChat
+
 LibreChat brings together the future of assistant AIs with the revolutionary technology of OpenAI's ChatGPT. Celebrating the original styling, LibreChat gives you the ability to integrate multiple AI models. It also integrates and enhances original client features such as conversation and message search, prompt templates and plugins.

 With LibreChat, you no longer need to opt for ChatGPT Plus and can instead use free or pay-per-call APIs. We welcome contributions, cloning, and forking to enhance the capabilities of this advanced chatbot platform.
-  
-<!-- https://github.com/danny-avila/LibreChat/assets/110412045/c1eb0c0f-41f6-4335-b982-84b278b53d59 -->

-[![Watch the video](https://img.youtube.com/vi/pNIOs1ovsXw/maxresdefault.jpg)](https://youtu.be/pNIOs1ovsXw)
+[![Watch the video](https://raw.githubusercontent.com/LibreChat-AI/librechat.ai/main/public/images/changelog/v0.7.4.png)](https://www.youtube.com/watch?v=cvosUxogdpI)
 Click on the thumbnail to open the video☝️

-# Features
- Response streaming identical to ChatGPT through server-sent events
- UI from original ChatGPT, including Dark mode
- AI model selection: OpenAI API, BingAI, ChatGPT Browser, PaLM2, Anthropic (Claude), Plugins
- Create, Save, & Share custom presets - [More info on prompt presets here](https://github.com/danny-avila/LibreChat/releases/tag/v0.3.0)
- Edit and Resubmit messages with conversation branching
- Search all messages/conversations - [More info here](https://github.com/danny-avila/LibreChat/releases/tag/v0.1.0)
- Plugins now available (including web access, image generation and more)
+---
+
+## 🌐 Resources
+
+**GitHub Repo:**
+  - **RAG API:** [github.com/danny-avila/rag_api](https://github.com/danny-avila/rag_api)
+  - **Website:** [github.com/LibreChat-AI/librechat.ai](https://github.com/LibreChat-AI/librechat.ai)
+
+**Other:**
+  - **Website:** [librechat.ai](https://librechat.ai)
+  - **Documentation:** [docs.librechat.ai](https://docs.librechat.ai)
+  - **Blog:** [blog.librechat.ai](https://docs.librechat.ai)

 ---

-## ⚠️ [Breaking Changes](docs/general_info/breaking_changes.md) ⚠️
+## 📝 Changelog

-**Please read this before updating from a previous version**
+Keep up with the latest updates by visiting the releases page and notes:
+- [Releases](https://github.com/danny-avila/LibreChat/releases)
+- [Changelog](https://www.librechat.ai/changelog) 
+
+**⚠️ Please consult the [changelog](https://www.librechat.ai/changelog) for breaking changes before updating.**

 ---

-## Changelog 
-Keep up with the latest updates by visiting the releases page - [Releases](https://github.com/danny-avila/LibreChat/releases)
+## ⭐ Star History
+
+<p align="center">
+  <a href="https://star-history.com/#danny-avila/LibreChat&Date">
+    <img alt="Star History Chart" src="https://api.star-history.com/svg?repos=danny-avila/LibreChat&type=Date&theme=dark" onerror="this.src='https://api.star-history.com/svg?repos=danny-avila/LibreChat&type=Date'" />
+  </a>
+</p>
+<p align="center">
+  <a href="https://trendshift.io/repositories/4685" target="_blank" style="padding: 10px;">
+    <img src="https://trendshift.io/api/badge/repositories/4685" alt="danny-avila%2FLibreChat | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/>
+  </a>
+  <a href="https://runacap.com/ross-index/q1-24/" target="_blank" rel="noopener" style="margin-left: 20px;">
+    <img style="width: 260px; height: 56px" src="https://runacap.com/wp-content/uploads/2024/04/ROSS_badge_white_Q1_2024.svg" alt="ROSS Index - Fastest Growing Open-Source Startups in Q1 2024 | Runa Capital" width="260" height="56"/>
+  </a>
+</p>

 ---

-<h1>Table of Contents</h1>
+## ✨ Contributions

-<details open>
-  <summary><strong>Getting Started</strong></summary>
-
-  * Installation
-    * [Docker Compose Install🐳](docs/install/docker_compose_install.md)
-    * [Linux Install🐧](docs/install/linux_install.md)
-    * [Mac Install🍎](docs/install/mac_install.md)
-    * [Windows Install💙](docs/install/windows_install.md)
-  * Configuration
-    * [APIs and Tokens](docs/install/apis_and_tokens.md)
-    * [User Auth System](docs/install/user_auth_system.md)
-    * [Online MongoDB Database](docs/install/mongodb.md)
-    * [Default Language](docs/install/default_language.md)
-</details>
-
-<details>
-  <summary><strong>General Information</strong></summary>
-
-  * [Code of Conduct](.github/CODE_OF_CONDUCT.md)
-  * [Project Origin](docs/general_info/project_origin.md)
-  * [Multilingual Information](docs/general_info/multilingual_information.md)
-  * [Tech Stack](docs/general_info/tech_stack.md)   
-</details>
-
-<details>
-  <summary><strong>Features</strong></summary>
-
-  * **Plugins**
-    * [Introduction](docs/features/plugins/introduction.md)
-    * [Google](docs/features/plugins/google_search.md)
-    * [Stable Diffusion](docs/features/plugins/stable_diffusion.md)
-    * [Wolfram](docs/features/plugins/wolfram.md)
-    * [Make Your Own Plugin](docs/features/plugins/make_your_own.md)
-    * [Using official ChatGPT Plugins](docs/features/plugins/chatgpt_plugins_openapi.md)
-
-  
-  * [Automated Moderation](docs/features/mod_system.md)
-  * [Third-Party Tools](docs/features/third_party.md)
-  * [Proxy](docs/features/proxy.md)
-  * [Bing Jailbreak](docs/features/bing_jailbreak.md)
-</details>
-
-<details>
-  <summary><strong>Cloud Deployment</strong></summary>
-
-  * [DigitalOcean](docs/deployment/digitalocean.md)
-  * [Azure](docs/deployment/azure-terraform.md)
-  * [Linode](docs/deployment/linode.md)
-  * [Cloudflare](docs/deployment/cloudflare.md)
-  * [Ngrok](docs/deployment/ngrok.md)
-  * [HuggingFace](docs/deployment/huggingface.md)
-  * [Render](docs/deployment/render.md)
-  * [Hetzner](docs/deployment/hetzner_ubuntu.md)
-  * [Heroku](docs/deployment/heroku.md)
-</details>
-
-<details>
-  <summary><strong>Contributions</strong></summary>
-  
-  * [Contributor Guidelines](.github/CONTRIBUTING.md)
-  * [Documentation Guidelines](docs/contributions/documentation_guidelines.md)
-  * [Contribute a Translation](docs/contributions/translation_contribution.md)
-  * [Code Standards and Conventions](docs/contributions/coding_conventions.md)
-  * [Testing](docs/contributions/testing.md)
-  * [Security](.github/SECURITY.md)
-  * [Project Roadmap](https://github.com/users/danny-avila/projects/2)
-</details>
+Contributions, suggestions, bug reports and fixes are welcome!

+For new features, components, or extensions, please open an issue and discuss before sending a PR.

 ---

-## Star History
+## 💖 This project exists in its current state thanks to all the people who contribute

-[![Star History Chart](https://api.star-history.com/svg?repos=danny-avila/LibreChat&type=Date)](https://star-history.com/#danny-avila/LibreChat&Date)
-
---
-
-## Sponsors
-
-  Sponsored by <a href="https://github.com/mjtechguy"><b>@mjtechguy</b></a>, <a href="https://github.com/SphaeroX"><b>@SphaeroX</b></a>, <a href="https://github.com/DavidDev1334"><b>@DavidDev1334</b></a>, <a href="https://github.com/fuegovic"><b>@fuegovic</b></a>, <a href="https://github.com/Pharrcyde"><b>@Pharrcyde</b></a> 
-  
---
-
-## Contributors
-Contributions and suggestions bug reports and fixes are welcome!
-Please read the documentation before you do!
-
---
-
-For new features, components, or extensions, please open an issue and discuss before sending a PR. 
-
- Join the [Discord community](https://discord.gg/uDyZ5Tzhct)
-
-This project exists in its current state thanks to all the people who contribute
---
 <a href="https://github.com/danny-avila/LibreChat/graphs/contributors">
  <img src="https://contrib.rocks/image?repo=danny-avila/LibreChat" />
 </a>
--- a/api/app/bingai.js
+++ b/api/app/bingai.js
@@ -1,6 +1,8 @@
 require('dotenv').config();
 const { KeyvFile } = require('keyv-file');
-const { getUserKey, checkUserKeyExpiry } = require('../server/services/UserService');
+const { EModelEndpoint } = require('librechat-data-provider');
+const { getUserKey, checkUserKeyExpiry } = require('~/server/services/UserService');
+const { logger } = require('~/config');

 const askBing = async ({
  text,
@@ -22,14 +24,11 @@ const askBing = async ({

  let key = null;
  if (expiresAt && isUserProvided) {
-    checkUserKeyExpiry(
-      expiresAt,
-      'Your BingAI Cookies have expired. Please provide your cookies again.',
-    );
+    checkUserKeyExpiry(expiresAt, EModelEndpoint.bingAI);
    key = await getUserKey({ userId, name: 'bingAI' });
  }

-  const { BingAIClient } = await import('@waylaidwanderer/chatgpt-api');
+  const { BingAIClient } = await import('nodejs-gpt');
  const store = {
    store: new KeyvFile({ filename: './data/cache.json' }),
  };
@@ -94,13 +93,13 @@ const askBing = async ({
    // don't give those parameters for new conversation
    // for new conversation, conversationSignature always is null
    if (conversationSignature) {
-      options.conversationSignature = conversationSignature;
+      options.encryptedConversationSignature = conversationSignature;
      options.clientId = clientId;
      options.invocationId = invocationId;
    }
  }

-  console.log('bing options', options);
+  logger.debug('bing options', options);

  const res = await bingAIClient.sendMessage(text, options);

--- a/api/app/chatgpt-browser.js
+++ b/api/app/chatgpt-browser.js
@@ -1,5 +1,6 @@
 require('dotenv').config();
 const { KeyvFile } = require('keyv-file');
+const { Constants, EModelEndpoint } = require('librechat-data-provider');
 const { getUserKey, checkUserKeyExpiry } = require('../server/services/UserService');

 const browserClient = async ({
@@ -17,14 +18,11 @@ const browserClient = async ({

  let key = null;
  if (expiresAt && isUserProvided) {
-    checkUserKeyExpiry(
-      expiresAt,
-      'Your ChatGPT Access Token has expired. Please provide your token again.',
-    );
+    checkUserKeyExpiry(expiresAt, EModelEndpoint.chatGPTBrowser);
    key = await getUserKey({ userId, name: 'chatGPTBrowser' });
  }

-  const { ChatGPTBrowserClient } = await import('@waylaidwanderer/chatgpt-api');
+  const { ChatGPTBrowserClient } = await import('nodejs-gpt');
  const store = {
    store: new KeyvFile({ filename: './data/cache.json' }),
  };
@@ -48,7 +46,7 @@ const browserClient = async ({
    options = { ...options, parentMessageId, conversationId };
  }

-  if (parentMessageId === '00000000-0000-0000-0000-000000000000') {
+  if (parentMessageId === Constants.NO_PARENT) {
    delete options.conversationId;
  }

--- a/api/app/clients/AnthropicClient.js
+++ b/api/app/clients/AnthropicClient.js
@@ -1,21 +1,69 @@
-// const { Agent, ProxyAgent } = require('undici');
-const BaseClient = require('./BaseClient');
-const { encoding_for_model: encodingForModel, get_encoding: getEncoding } = require('tiktoken');
 const Anthropic = require('@anthropic-ai/sdk');
+const { HttpsProxyAgent } = require('https-proxy-agent');
+const { encoding_for_model: encodingForModel, get_encoding: getEncoding } = require('tiktoken');
+const {
+  Constants,
+  EModelEndpoint,
+  anthropicSettings,
+  getResponseSender,
+  validateVisionModel,
+} = require('librechat-data-provider');
+const { encodeAndFormat } = require('~/server/services/Files/images/encode');
+const {
+  truncateText,
+  formatMessage,
+  addCacheControl,
+  titleFunctionPrompt,
+  parseParamFromPrompt,
+  createContextHandlers,
+} = require('./prompts');
+const { spendTokens, spendStructuredTokens } = require('~/models/spendTokens');
+const { getModelMaxTokens, matchModelName } = require('~/utils');
+const { sleep } = require('~/server/utils');
+const BaseClient = require('./BaseClient');
+const { logger } = require('~/config');

 const HUMAN_PROMPT = '\n\nHuman:';
 const AI_PROMPT = '\n\nAssistant:';

 const tokenizersCache = {};

+/** Helper function to introduce a delay before retrying */
+function delayBeforeRetry(attempts, baseDelay = 1000) {
+  return new Promise((resolve) => setTimeout(resolve, baseDelay * attempts));
+}
+
+const tokenEventTypes = new Set(['message_start', 'message_delta']);
+const { legacy } = anthropicSettings;
+
 class AnthropicClient extends BaseClient {
-  constructor(apiKey, options = {}, cacheOptions = {}) {
-    super(apiKey, options, cacheOptions);
+  constructor(apiKey, options = {}) {
+    super(apiKey, options);
    this.apiKey = apiKey || process.env.ANTHROPIC_API_KEY;
-    this.sender = 'Anthropic';
    this.userLabel = HUMAN_PROMPT;
    this.assistantLabel = AI_PROMPT;
+    this.contextStrategy = options.contextStrategy
+      ? options.contextStrategy.toLowerCase()
+      : 'discard';
    this.setOptions(options);
+    /** @type {string | undefined} */
+    this.systemMessage;
+    /** @type {AnthropicMessageStartEvent| undefined} */
+    this.message_start;
+    /** @type {AnthropicMessageDeltaEvent| undefined} */
+    this.message_delta;
+    /** Whether the model is part of the Claude 3 Family
+     * @type {boolean} */
+    this.isClaude3;
+    /** 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;
  }

  setOptions(options) {
@@ -35,18 +83,37 @@ class AnthropicClient extends BaseClient {
      this.options = options;
    }

-    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 || 'claude-1',
-      temperature: typeof modelOptions.temperature === 'undefined' ? 0.7 : modelOptions.temperature, // 0 - 1, 0.7 is recommended
-      topP: typeof modelOptions.topP === 'undefined' ? 0.7 : modelOptions.topP, // 0 - 1, default: 0.7
-      topK: typeof modelOptions.topK === 'undefined' ? 40 : modelOptions.topK, // 1-40, default: 40
-      stop: modelOptions.stop, // no stop method for now
-    };
+    this.modelOptions = Object.assign(
+      {
+        model: anthropicSettings.model.default,
+      },
+      this.modelOptions,
+      this.options.modelOptions,
+    );

-    this.maxContextTokens = this.options.maxContextTokens || 99999;
+    const modelMatch = matchModelName(this.modelOptions.model, EModelEndpoint.anthropic);
+    this.isClaude3 = modelMatch.startsWith('claude-3');
+    this.isLegacyOutput = !modelMatch.startsWith('claude-3-5-sonnet');
+    this.supportsCacheControl =
+      this.options.promptCache && this.checkPromptCacheSupport(modelMatch);
+
+    if (
+      this.isLegacyOutput &&
+      this.modelOptions.maxOutputTokens &&
+      this.modelOptions.maxOutputTokens > legacy.maxOutputTokens.default
+    ) {
+      this.modelOptions.maxOutputTokens = legacy.maxOutputTokens.default;
+    }
+
+    this.useMessages = this.isClaude3 || !!this.options.attachments;
+
+    this.defaultVisionModel = this.options.visionModel ?? 'claude-3-sonnet-20240229';
+    this.options.attachments?.then((attachments) => this.checkVisionRequest(attachments));
+
+    this.maxContextTokens =
+      this.options.maxContextTokens ??
+      getModelMaxTokens(this.modelOptions.model, EModelEndpoint.anthropic) ??
+      100000;
    this.maxResponseTokens = this.modelOptions.maxOutputTokens || 1500;
    this.maxPromptTokens =
      this.options.maxPromptTokens || this.maxContextTokens - this.maxResponseTokens;
@@ -59,6 +126,14 @@ class AnthropicClient extends BaseClient {
      );
    }

+    this.sender =
+      this.options.sender ??
+      getResponseSender({
+        model: this.modelOptions.model,
+        endpoint: EModelEndpoint.anthropic,
+        modelLabel: this.options.modelLabel,
+      });
+
    this.startToken = '||>';
    this.endToken = '';
    this.gptEncoder = this.constructor.getTokenizer('cl100k_base');
@@ -77,47 +152,337 @@ class AnthropicClient extends BaseClient {
    return this;
  }

-  getClient() {
+  /**
+   * Get the initialized Anthropic client.
+   * @param {Partial<Anthropic.ClientOptions>} requestOptions - The options for the client.
+   * @returns {Anthropic} The Anthropic client instance.
+   */
+  getClient(requestOptions) {
+    /** @type {Anthropic.ClientOptions} */
+    const options = {
+      fetch: this.fetch,
+      apiKey: this.apiKey,
+    };
+
+    if (this.options.proxy) {
+      options.httpAgent = new HttpsProxyAgent(this.options.proxy);
+    }
+
    if (this.options.reverseProxyUrl) {
-      return new Anthropic({
-        apiKey: this.apiKey,
-        baseURL: this.options.reverseProxyUrl,
-      });
-    } else {
-      return new Anthropic({
-        apiKey: this.apiKey,
-      });
+      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',
+      };
+    }
+
+    return new Anthropic(options);
+  }
+
+  /**
+   * Get stream usage as returned by this client's API response.
+   * @returns {AnthropicStreamUsage} The stream usage object.
+   */
+  getStreamUsage() {
+    const inputUsage = this.message_start?.message?.usage ?? {};
+    const outputUsage = this.message_delta?.usage ?? {};
+    return Object.assign({}, inputUsage, outputUsage);
+  }
+
+  /**
+   * Calculates the correct token count for the current 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.
+   * If revisiting a conversation with a chat history entirely composed of token estimates,
+   * the cumulative token count going forward should become more accurate as the conversation progresses.
+   * @param {Object} params - The parameters for the calculation.
+   * @param {Record<string, number>} params.tokenCountMap - A map of message IDs to their token counts.
+   * @param {string} params.currentMessageId - The ID of the current message to calculate.
+   * @param {AnthropicStreamUsage} params.usage - The usage object returned by the API.
+   * @returns {number} The correct token count for the current message.
+   */
+  calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage }) {
+    const originalEstimate = tokenCountMap[currentMessageId] || 0;
+
+    if (!usage || typeof usage.input_tokens !== 'number') {
+      return originalEstimate;
+    }
+
+    tokenCountMap[currentMessageId] = 0;
+    const totalTokensFromMap = Object.values(tokenCountMap).reduce((sum, count) => {
+      const numCount = Number(count);
+      return sum + (isNaN(numCount) ? 0 : numCount);
+    }, 0);
+    const totalInputTokens =
+      (usage.input_tokens ?? 0) +
+      (usage.cache_creation_input_tokens ?? 0) +
+      (usage.cache_read_input_tokens ?? 0);
+
+    const currentMessageTokens = totalInputTokens - totalTokensFromMap;
+    return currentMessageTokens > 0 ? currentMessageTokens : originalEstimate;
+  }
+
+  /**
+   * Get Token Count for LibreChat Message
+   * @param {TMessage} responseMessage
+   * @returns {number}
+   */
+  getTokenCountForResponse(responseMessage) {
+    return this.getTokenCountForMessage({
+      role: 'assistant',
+      content: responseMessage.text,
+    });
+  }
+
+  /**
+   *
+   * Checks if the model is a vision model based on request attachments and sets the appropriate options:
+   * - Sets `this.modelOptions.model` to `gpt-4-vision-preview` if the request is a vision request.
+   * - Sets `this.isVisionModel` to `true` if vision request.
+   * - Deletes `this.modelOptions.stop` if vision request.
+   * @param {MongoFile[]} attachments
+   */
+  checkVisionRequest(attachments) {
+    const availableModels = this.options.modelsConfig?.[EModelEndpoint.anthropic];
+    this.isVisionModel = validateVisionModel({ model: this.modelOptions.model, availableModels });
+
+    const visionModelAvailable = availableModels?.includes(this.defaultVisionModel);
+    if (
+      attachments &&
+      attachments.some((file) => file?.type && file?.type?.includes('image')) &&
+      visionModelAvailable &&
+      !this.isVisionModel
+    ) {
+      this.modelOptions.model = this.defaultVisionModel;
+      this.isVisionModel = true;
    }
  }

-  async buildMessages(messages, parentMessageId) {
-    const orderedMessages = this.constructor.getMessagesForConversation(messages, parentMessageId);
-    if (this.options.debug) {
-      console.debug('AnthropicClient: orderedMessages', orderedMessages, parentMessageId);
+  /**
+   * Calculate the token cost in tokens for an image based on its dimensions and detail level.
+   *
+   * For reference, see: https://docs.anthropic.com/claude/docs/vision#image-costs
+   *
+   * @param {Object} image - The image object.
+   * @param {number} image.width - The width of the image.
+   * @param {number} image.height - The height of the image.
+   * @returns {number} The calculated token cost measured by tokens.
+   *
+   */
+  calculateImageTokenCost({ width, height }) {
+    return Math.ceil((width * height) / 750);
+  }
+
+  async addImageURLs(message, attachments) {
+    const { files, image_urls } = await encodeAndFormat(
+      this.options.req,
+      attachments,
+      EModelEndpoint.anthropic,
+    );
+    message.image_urls = image_urls.length ? image_urls : undefined;
+    return files;
+  }
+
+  /**
+   * @param {object} params
+   * @param {number} params.promptTokens
+   * @param {number} params.completionTokens
+   * @param {AnthropicStreamUsage} [params.usage]
+   * @param {string} [params.model]
+   * @param {string} [params.context='message']
+   * @returns {Promise<void>}
+   */
+  async recordTokenUsage({ promptTokens, completionTokens, usage, model, context = 'message' }) {
+    if (usage != null && usage?.input_tokens != null) {
+      const input = usage.input_tokens ?? 0;
+      const write = usage.cache_creation_input_tokens ?? 0;
+      const read = usage.cache_read_input_tokens ?? 0;
+
+      await spendStructuredTokens(
+        {
+          context,
+          user: this.user,
+          conversationId: this.conversationId,
+          model: model ?? this.modelOptions.model,
+          endpointTokenConfig: this.options.endpointTokenConfig,
+        },
+        {
+          promptTokens: { input, write, read },
+          completionTokens,
+        },
+      );
+
+      return;
    }

-    const formattedMessages = orderedMessages.map((message) => ({
-      author: message.isCreatedByUser ? this.userLabel : this.assistantLabel,
-      content: message?.content ?? message.text,
-    }));
+    await spendTokens(
+      {
+        context,
+        user: this.user,
+        conversationId: this.conversationId,
+        model: model ?? this.modelOptions.model,
+        endpointTokenConfig: this.options.endpointTokenConfig,
+      },
+      { promptTokens, completionTokens },
+    );
+  }
+
+  async buildMessages(messages, parentMessageId) {
+    const orderedMessages = this.constructor.getMessagesForConversation({
+      messages,
+      parentMessageId,
+    });
+
+    logger.debug('[AnthropicClient] orderedMessages', { orderedMessages, parentMessageId });
+
+    if (this.options.attachments) {
+      const attachments = await this.options.attachments;
+      const images = attachments.filter((file) => file.type.includes('image'));
+
+      if (images.length && !this.isVisionModel) {
+        throw new Error('Images are only supported with the Claude 3 family of models');
+      }
+
+      const latestMessage = orderedMessages[orderedMessages.length - 1];
+
+      if (this.message_file_map) {
+        this.message_file_map[latestMessage.messageId] = attachments;
+      } else {
+        this.message_file_map = {
+          [latestMessage.messageId]: attachments,
+        };
+      }
+
+      const files = await this.addImageURLs(latestMessage, attachments);
+
+      this.options.attachments = files;
+    }
+
+    if (this.message_file_map) {
+      this.contextHandlers = createContextHandlers(
+        this.options.req,
+        orderedMessages[orderedMessages.length - 1].text,
+      );
+    }
+
+    const formattedMessages = orderedMessages.map((message, i) => {
+      const formattedMessage = this.useMessages
+        ? formatMessage({
+          message,
+          endpoint: EModelEndpoint.anthropic,
+        })
+        : {
+          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 */
+      if (needsTokenCount || (this.isVisionModel && (message.image_urls || message.files))) {
+        orderedMessages[i].tokenCount = this.getTokenCountForMessage(formattedMessage);
+      }
+
+      /* If message has files, calculate image token cost */
+      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;
+          }
+
+          orderedMessages[i].tokenCount += this.calculateImageTokenCost({
+            width: file.width,
+            height: file.height,
+          });
+        }
+      }
+
+      formattedMessage.tokenCount = orderedMessages[i].tokenCount;
+      return formattedMessage;
+    });
+
+    if (this.contextHandlers) {
+      this.augmentedPrompt = await this.contextHandlers.createContext();
+      this.options.promptPrefix = this.augmentedPrompt + (this.options.promptPrefix ?? '');
+    }
+
+    let { context: messagesInWindow, remainingContextTokens } =
+      await this.getMessagesWithinTokenLimit(formattedMessages);
+
+    const tokenCountMap = orderedMessages
+      .slice(orderedMessages.length - messagesInWindow.length)
+      .reduce((map, message, index) => {
+        const { messageId } = message;
+        if (!messageId) {
+          return map;
+        }
+
+        map[messageId] = orderedMessages[index].tokenCount;
+        return map;
+      }, {});
+
+    logger.debug('[AnthropicClient]', {
+      messagesInWindow: messagesInWindow.length,
+      remainingContextTokens,
+    });

    let lastAuthor = '';
    let groupedMessages = [];

-    for (let message of formattedMessages) {
+    for (let i = 0; i < messagesInWindow.length; i++) {
+      const message = messagesInWindow[i];
+      const author = message.role ?? message.author;
      // If last author is not same as current author, add to new group
-      if (lastAuthor !== message.author) {
-        groupedMessages.push({
-          author: message.author,
+      if (lastAuthor !== author) {
+        const newMessage = {
          content: [message.content],
-        });
-        lastAuthor = message.author;
+        };
+
+        if (message.role) {
+          newMessage.role = message.role;
+        } else {
+          newMessage.author = message.author;
+        }
+
+        groupedMessages.push(newMessage);
+        lastAuthor = author;
        // If same author, append content to the last group
      } else {
        groupedMessages[groupedMessages.length - 1].content.push(message.content);
      }
    }

+    groupedMessages = groupedMessages.map((msg, i) => {
+      const isLast = i === groupedMessages.length - 1;
+      if (msg.content.length === 1) {
+        const content = msg.content[0];
+        return {
+          ...msg,
+          // reason: final assistant content cannot end with trailing whitespace
+          content:
+            isLast && this.useMessages && msg.role === 'assistant' && typeof content === 'string'
+              ? content?.trim()
+              : content,
+        };
+      }
+
+      if (!this.useMessages && msg.tokenCount) {
+        delete msg.tokenCount;
+      }
+
+      return msg;
+    });
+
    let identityPrefix = '';
    if (this.options.userLabel) {
      identityPrefix = `\nHuman's name: ${this.options.userLabel}`;
@@ -127,7 +492,10 @@ class AnthropicClient extends BaseClient {
      identityPrefix = `${identityPrefix}\nYou are ${this.options.modelLabel}`;
    }

-    let promptPrefix = (this.options.promptPrefix || '').trim();
+    let promptPrefix = (this.options.promptPrefix ?? '').trim();
+    if (typeof this.options.artifactsPrompt === 'string' && this.options.artifactsPrompt) {
+      promptPrefix = `${promptPrefix ?? ''}\n${this.options.artifactsPrompt}`.trim();
+    }
    if (promptPrefix) {
      // If the prompt prefix doesn't end with the end token, add it.
      if (!promptPrefix.endsWith(`${this.endToken}`)) {
@@ -143,9 +511,10 @@ class AnthropicClient extends BaseClient {
    // Prompt AI to respond, empty if last message was from AI
    let isEdited = lastAuthor === this.assistantLabel;
    const promptSuffix = isEdited ? '' : `${promptPrefix}${this.assistantLabel}\n`;
-    let currentTokenCount = isEdited
-      ? this.getTokenCount(promptPrefix)
-      : this.getTokenCount(promptSuffix);
+    let currentTokenCount =
+      isEdited || this.useMessages
+        ? this.getTokenCount(promptPrefix)
+        : this.getTokenCount(promptSuffix);

    let promptBody = '';
    const maxTokenCount = this.maxPromptTokens;
@@ -213,7 +582,69 @@ class AnthropicClient extends BaseClient {
      return true;
    };

-    await buildPromptBody();
+    const messagesPayload = [];
+    const buildMessagesPayload = async () => {
+      let canContinue = true;
+
+      if (promptPrefix) {
+        this.systemMessage = promptPrefix;
+      }
+
+      while (currentTokenCount < maxTokenCount && groupedMessages.length > 0 && canContinue) {
+        const message = groupedMessages.pop();
+
+        let tokenCountForMessage = message.tokenCount ?? this.getTokenCountForMessage(message);
+
+        const newTokenCount = currentTokenCount + tokenCountForMessage;
+        const exceededMaxCount = newTokenCount > maxTokenCount;
+
+        if (exceededMaxCount && messagesPayload.length === 0) {
+          throw new Error(
+            `Prompt is too long. Max token count is ${maxTokenCount}, but prompt is ${newTokenCount} tokens long.`,
+          );
+        } else if (exceededMaxCount) {
+          canContinue = false;
+          break;
+        }
+
+        delete message.tokenCount;
+        messagesPayload.unshift(message);
+        currentTokenCount = newTokenCount;
+
+        // Switch off isEdited after using it once
+        if (isEdited && message.role === 'assistant') {
+          isEdited = false;
+        }
+
+        // Wait for next tick to avoid blocking the event loop
+        await new Promise((resolve) => setImmediate(resolve));
+      }
+    };
+
+    const processTokens = () => {
+      // Add 2 tokens for metadata after all messages have been counted.
+      currentTokenCount += 2;
+
+      // Use up to `this.maxContextTokens` tokens (prompt + response), but try to leave `this.maxTokens` tokens for the response.
+      this.modelOptions.maxOutputTokens = Math.min(
+        this.maxContextTokens - currentTokenCount,
+        this.maxResponseTokens,
+      );
+    };
+
+    if (this.modelOptions.model.startsWith('claude-3')) {
+      await buildMessagesPayload();
+      processTokens();
+      return {
+        prompt: messagesPayload,
+        context: messagesInWindow,
+        promptTokens: currentTokenCount,
+        tokenCountMap,
+      };
+    } else {
+      await buildPromptBody();
+      processTokens();
+    }

    if (nextMessage.remove) {
      promptBody = promptBody.replace(nextMessage.messageString, '');
@@ -223,23 +654,38 @@ class AnthropicClient extends BaseClient {

    let prompt = `${promptBody}${promptSuffix}`;

-    // Add 2 tokens for metadata after all messages have been counted.
-    currentTokenCount += 2;
-
-    // Use up to `this.maxContextTokens` tokens (prompt + response), but try to leave `this.maxTokens` tokens for the response.
-    this.modelOptions.maxOutputTokens = Math.min(
-      this.maxContextTokens - currentTokenCount,
-      this.maxResponseTokens,
-    );
-
-    return { prompt, context };
+    return { prompt, context, promptTokens: currentTokenCount, tokenCountMap };
  }

  getCompletion() {
-    console.log('AnthropicClient doesn\'t use getCompletion (all handled in sendCompletion)');
+    logger.debug('AnthropicClient doesn\'t use getCompletion (all handled in sendCompletion)');
+  }
+
+  /**
+   * Creates a message or completion response using the Anthropic client.
+   * @param {Anthropic} client - The Anthropic client instance.
+   * @param {Anthropic.default.MessageCreateParams | Anthropic.default.CompletionCreateParams} options - The options for the message or completion.
+   * @param {boolean} useMessages - Whether to use messages or completions. Defaults to `this.useMessages`.
+   * @returns {Promise<Anthropic.default.Message | Anthropic.default.Completion>} The response from the Anthropic client.
+   */
+  async createResponse(client, options, useMessages) {
+    return useMessages ?? this.useMessages
+      ? await client.messages.create(options)
+      : await client.completions.create(options);
+  }
+
+  /**
+   * @param {string} modelName
+   * @returns {boolean}
+   */
+  checkPromptCacheSupport(modelName) {
+    const modelMatch = matchModelName(modelName, EModelEndpoint.anthropic);
+    if (modelMatch === 'claude-3-5-sonnet' || modelMatch === 'claude-3-haiku') {
+      return true;
+    }
+    return false;
  }

-  // TODO: implement abortController usage
  async sendCompletion(payload, { onProgress, abortController }) {
    if (!abortController) {
      abortController = new AbortController();
@@ -252,14 +698,7 @@ class AnthropicClient extends BaseClient {
      modelOptions.stream = true;
    }

-    const { debug } = this.options;
-    if (debug) {
-      console.debug();
-      console.debug(modelOptions);
-      console.debug();
-    }
-
-    const client = this.getClient();
+    logger.debug('modelOptions', { modelOptions });
    const metadata = {
      user_id: this.user,
    };
@@ -274,69 +713,130 @@ class AnthropicClient extends BaseClient {
      topP: top_p,
      topK: top_k,
    } = this.modelOptions;
+
    const requestOptions = {
-      prompt: payload,
      model,
      stream: stream || true,
-      max_tokens_to_sample: maxOutputTokens || 1500,
      stop_sequences,
      temperature,
      metadata,
      top_p,
      top_k,
    };
-    if (this.options.debug) {
-      console.log('AnthropicClient: requestOptions');
-      console.dir(requestOptions, { depth: null });
-    }
-    const response = await client.completions.create(requestOptions);

-    signal.addEventListener('abort', () => {
-      if (this.options.debug) {
-        console.log('AnthropicClient: message aborted!');
-      }
-      response.controller.abort();
-    });
-
-    for await (const completion of response) {
-      if (this.options.debug) {
-        // Uncomment to debug message stream
-        // console.debug(completion);
-      }
-      text += completion.completion;
-      onProgress(completion.completion);
+    if (this.useMessages) {
+      requestOptions.messages = payload;
+      requestOptions.max_tokens = maxOutputTokens || legacy.maxOutputTokens.default;
+    } else {
+      requestOptions.prompt = payload;
+      requestOptions.max_tokens_to_sample = maxOutputTokens || 1500;
    }

-    signal.removeEventListener('abort', () => {
-      if (this.options.debug) {
-        console.log('AnthropicClient: message aborted!');
+    if (this.systemMessage && this.supportsCacheControl === true) {
+      requestOptions.system = [
+        {
+          type: 'text',
+          text: this.systemMessage,
+          cache_control: { type: 'ephemeral' },
+        },
+      ];
+    } else if (this.systemMessage) {
+      requestOptions.system = this.systemMessage;
+    }
+
+    if (this.supportsCacheControl === true && this.useMessages) {
+      requestOptions.messages = addCacheControl(requestOptions.messages);
+    }
+
+    logger.debug('[AnthropicClient]', { ...requestOptions });
+
+    const handleChunk = (currentChunk) => {
+      if (currentChunk) {
+        text += currentChunk;
+        onProgress(currentChunk);
      }
-      response.controller.abort();
-    });
+    };
+
+    const maxRetries = 3;
+    const streamRate = this.options.streamRate ?? Constants.DEFAULT_STREAM_RATE;
+    async function processResponse() {
+      let attempts = 0;
+
+      while (attempts < maxRetries) {
+        let response;
+        try {
+          const client = this.getClient(requestOptions);
+          response = await this.createResponse(client, requestOptions);
+
+          signal.addEventListener('abort', () => {
+            logger.debug('[AnthropicClient] message aborted!');
+            if (response.controller?.abort) {
+              response.controller.abort();
+            }
+          });
+
+          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);
+            }
+
+            await sleep(streamRate);
+          }
+
+          // Successful processing, exit loop
+          break;
+        } catch (error) {
+          attempts += 1;
+          logger.warn(
+            `User: ${this.user} | Anthropic Request ${attempts} failed: ${error.message}`,
+          );
+
+          if (attempts < maxRetries) {
+            await delayBeforeRetry(attempts, 350);
+          } else {
+            throw new Error(`Operation failed after ${maxRetries} attempts: ${error.message}`);
+          }
+        } finally {
+          signal.removeEventListener('abort', () => {
+            logger.debug('[AnthropicClient] message aborted!');
+            if (response.controller?.abort) {
+              response.controller.abort();
+            }
+          });
+        }
+      }
+    }
+
+    await processResponse.bind(this)();

    return text.trim();
  }

-  // I commented this out because I will need to refactor this for the BaseClient/all clients
-  // getMessageMapMethod() {
-  //   return ((message) => ({
-  //     author: message.isCreatedByUser ? this.userLabel : this.assistantLabel,
-  //     content: message?.content ?? message.text
-  //   })).bind(this);
-  // }
-
  getSaveOptions() {
    return {
+      maxContextTokens: this.options.maxContextTokens,
+      artifacts: this.options.artifacts,
      promptPrefix: this.options.promptPrefix,
      modelLabel: this.options.modelLabel,
+      promptCache: this.options.promptCache,
+      resendFiles: this.options.resendFiles,
+      iconURL: this.options.iconURL,
+      greeting: this.options.greeting,
+      spec: this.options.spec,
      ...this.modelOptions,
    };
  }

  getBuildMessagesOptions() {
-    if (this.options.debug) {
-      console.log('AnthropicClient doesn\'t use getBuildMessagesOptions');
-    }
+    logger.debug('AnthropicClient doesn\'t use getBuildMessagesOptions');
  }

  static getTokenizer(encoding, isModelName = false, extendSpecialTokens = {}) {
@@ -356,6 +856,84 @@ class AnthropicClient extends BaseClient {
  getTokenCount(text) {
    return this.gptEncoder.encode(text, 'all').length;
  }
+
+  /**
+   * Generates a concise title for a conversation based on the user's input text and response.
+   * Involves sending a chat completion request with specific instructions for title generation.
+   *
+   * This function capitlizes on [Anthropic's function calling training](https://docs.anthropic.com/claude/docs/functions-external-tools).
+   *
+   * @param {Object} params - The parameters for the conversation title generation.
+   * @param {string} params.text - The user's input.
+   * @param {string} [params.responseText=''] - The AI's immediate response to the user.
+   *
+   * @returns {Promise<string | 'New Chat'>} A promise that resolves to the generated conversation title.
+   *                            In case of failure, it will return the default title, "New Chat".
+   */
+  async titleConvo({ text, responseText = '' }) {
+    let title = 'New Chat';
+    this.message_delta = undefined;
+    this.message_start = undefined;
+    const convo = `<initial_message>
+  ${truncateText(text)}
+  </initial_message>
+  <response>
+  ${JSON.stringify(truncateText(responseText))}
+  </response>`;
+
+    const { ANTHROPIC_TITLE_MODEL } = process.env ?? {};
+    const model = this.options.titleModel ?? ANTHROPIC_TITLE_MODEL ?? 'claude-3-haiku-20240307';
+    const system = titleFunctionPrompt;
+
+    const titleChatCompletion = async () => {
+      const content = `<conversation_context>
+  ${convo}
+  </conversation_context>
+  
+  Please generate a title for this conversation.`;
+
+      const titleMessage = { role: 'user', content };
+      const requestOptions = {
+        model,
+        temperature: 0.3,
+        max_tokens: 1024,
+        system,
+        stop_sequences: ['\n\nHuman:', '\n\nAssistant', '</function_calls>'],
+        messages: [titleMessage],
+      };
+
+      try {
+        const response = await this.createResponse(
+          this.getClient(requestOptions),
+          requestOptions,
+          true,
+        );
+        let promptTokens = response?.usage?.input_tokens;
+        let completionTokens = response?.usage?.output_tokens;
+        if (!promptTokens) {
+          promptTokens = this.getTokenCountForMessage(titleMessage);
+          promptTokens += this.getTokenCountForMessage({ role: 'system', content: system });
+        }
+        if (!completionTokens) {
+          completionTokens = this.getTokenCountForMessage(response.content[0]);
+        }
+        await this.recordTokenUsage({
+          model,
+          promptTokens,
+          completionTokens,
+          context: 'title',
+        });
+        const text = response.content[0].text;
+        title = parseParamFromPrompt(text, 'title');
+      } catch (e) {
+        logger.error('[AnthropicClient] There was an issue generating the title', e);
+      }
+    };
+
+    await titleChatCompletion();
+    logger.debug('[AnthropicClient] Convo Title: ' + title);
+    return title;
+  }
 }

 module.exports = AnthropicClient;
--- a/api/app/clients/BaseClient.js
+++ b/api/app/clients/BaseClient.js
--- a/api/app/clients/ChatGPTClient.js
+++ b/api/app/clients/ChatGPTClient.js
@@ -1,9 +1,19 @@
-const crypto = require('crypto');
 const Keyv = require('keyv');
+const crypto = require('crypto');
+const {
+  EModelEndpoint,
+  resolveHeaders,
+  CohereConstants,
+  mapModelToAzureConfig,
+} = require('librechat-data-provider');
+const { CohereClient } = require('cohere-ai');
 const { encoding_for_model: encodingForModel, get_encoding: getEncoding } = require('tiktoken');
 const { fetchEventSource } = require('@waylaidwanderer/fetch-event-source');
+const { createCoherePayload } = require('./llm');
 const { Agent, ProxyAgent } = require('undici');
 const BaseClient = require('./BaseClient');
+const { logger } = require('~/config');
+const { extractBaseURL, constructAzureURL, genAzureChatCompletion } = require('~/utils');

 const CHATGPT_MODEL = 'gpt-3.5-turbo';
 const tokenizersCache = {};
@@ -50,7 +60,7 @@ class ChatGPTClient extends BaseClient {
      stop: modelOptions.stop,
    };

-    this.isChatGptModel = this.modelOptions.model.startsWith('gpt-');
+    this.isChatGptModel = this.modelOptions.model.includes('gpt-');
    const { isChatGptModel } = this;
    this.isUnofficialChatGptModel =
      this.modelOptions.model.startsWith('text-chat') ||
@@ -140,11 +150,13 @@ class ChatGPTClient extends BaseClient {
    return tokenizer;
  }

-  async getCompletion(input, onProgress, abortController = null) {
+  /** @type {getCompletion} */
+  async getCompletion(input, onProgress, onTokenProgress, abortController = null) {
    if (!abortController) {
      abortController = new AbortController();
    }
-    const modelOptions = { ...this.modelOptions };
+
+    let modelOptions = { ...this.modelOptions };
    if (typeof onProgress === 'function') {
      modelOptions.stream = true;
    }
@@ -153,51 +165,182 @@ class ChatGPTClient extends BaseClient {
    } else {
      modelOptions.prompt = input;
    }
+
+    if (this.useOpenRouter && modelOptions.prompt) {
+      delete modelOptions.stop;
+    }
+
    const { debug } = this.options;
-    const url = this.completionsUrl;
+    let baseURL = this.completionsUrl;
    if (debug) {
      console.debug();
-      console.debug(url);
+      console.debug(baseURL);
      console.debug(modelOptions);
      console.debug();
    }
+
    const opts = {
      method: 'POST',
      headers: {
        'Content-Type': 'application/json',
      },
-      body: JSON.stringify(modelOptions),
      dispatcher: new Agent({
        bodyTimeout: 0,
        headersTimeout: 0,
      }),
    };

-    if (this.apiKey && this.options.azure) {
-      opts.headers['api-key'] = this.apiKey;
+    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 (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';
    }

-    if (this.options.headers) {
-      opts.headers = { ...opts.headers, ...this.options.headers };
-    }
-
    if (this.options.proxy) {
      opts.dispatcher = new ProxyAgent(this.options.proxy);
    }

+    /* 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(url, {
+          await fetchEventSource(completionsURL, {
            ...opts,
            signal: abortController.signal,
            async onopen(response) {
@@ -225,7 +368,6 @@ class ChatGPTClient extends BaseClient {
              // workaround for private API not sending [DONE] event
              if (!done) {
                onProgress('[DONE]');
-                abortController.abort();
                resolve();
              }
            },
@@ -238,14 +380,13 @@ class ChatGPTClient extends BaseClient {
            },
            onmessage(message) {
              if (debug) {
-                // console.debug(message);
+                console.debug(message);
              }
              if (!message.data || message.event === 'ping') {
                return;
              }
              if (message.data === '[DONE]') {
                onProgress('[DONE]');
-                abortController.abort();
                resolve();
                done = true;
                return;
@@ -258,7 +399,7 @@ class ChatGPTClient extends BaseClient {
        }
      });
    }
-    const response = await fetch(url, {
+    const response = await fetch(completionsURL, {
      ...opts,
      signal: abortController.signal,
    });
@@ -276,6 +417,43 @@ class ChatGPTClient extends BaseClient {
    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',
@@ -432,9 +610,7 @@ ${botMessage.message}
    return returnData;
  }

-  async buildPrompt(messages, parentMessageId, { isChatGptModel = false, promptPrefix = null }) {
-    const orderedMessages = this.constructor.getMessagesForConversation(messages, parentMessageId);
-
+  async buildPrompt(messages, { isChatGptModel = false, promptPrefix = null }) {
    promptPrefix = (promptPrefix || this.options.promptPrefix || '').trim();
    if (promptPrefix) {
      // If the prompt prefix doesn't end with the end token, add it.
@@ -480,8 +656,8 @@ ${botMessage.message}
    // 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();
+      if (currentTokenCount < maxTokenCount && messages.length > 0) {
+        const message = messages.pop();
        const roleLabel =
          message?.isCreatedByUser || message?.role?.toLowerCase() === 'user'
            ? this.userLabel
@@ -545,7 +721,7 @@ ${botMessage.message}
    if (isChatGptModel) {
      return { prompt: [instructionsPayload, messagePayload], context };
    }
-    return { prompt, context };
+    return { prompt, context, promptTokens: currentTokenCount };
  }

  getTokenCount(text) {
--- a/api/app/clients/GoogleClient.js
+++ b/api/app/clients/GoogleClient.js
@@ -1,23 +1,71 @@
-const BaseClient = require('./BaseClient');
 const { google } = require('googleapis');
 const { Agent, ProxyAgent } = require('undici');
+const { ChatVertexAI } = require('@langchain/google-vertexai');
+const { ChatGoogleGenerativeAI } = require('@langchain/google-genai');
+const { GoogleGenerativeAI: GenAI } = require('@google/generative-ai');
+const { GoogleVertexAI } = require('@langchain/community/llms/googlevertexai');
+const { ChatGoogleVertexAI } = require('langchain/chat_models/googlevertexai');
+const { AIMessage, HumanMessage, SystemMessage } = require('langchain/schema');
 const { encoding_for_model: encodingForModel, get_encoding: getEncoding } = require('tiktoken');
+const {
+  validateVisionModel,
+  getResponseSender,
+  endpointSettings,
+  EModelEndpoint,
+  VisionModes,
+  Constants,
+  AuthKeys,
+} = require('librechat-data-provider');
+const { encodeAndFormat } = require('~/server/services/Files/images');
+const { getModelMaxTokens } = require('~/utils');
+const { sleep } = require('~/server/utils');
+const { logger } = require('~/config');
+const {
+  formatMessage,
+  createContextHandlers,
+  titleInstruction,
+  truncateText,
+} = require('./prompts');
+const BaseClient = require('./BaseClient');

+const loc = 'us-central1';
+const publisher = 'google';
+const endpointPrefix = `https://${loc}-aiplatform.googleapis.com`;
+// const apiEndpoint = loc + '-aiplatform.googleapis.com';
 const tokenizersCache = {};

+const settings = endpointSettings[EModelEndpoint.google];
+
 class GoogleClient extends BaseClient {
  constructor(credentials, options = {}) {
    super('apiKey', options);
-    this.client_email = credentials.client_email;
-    this.project_id = credentials.project_id;
-    this.private_key = credentials.private_key;
-    this.sender = 'PaLM2';
+    let creds = {};
+
+    if (typeof credentials === 'string') {
+      creds = JSON.parse(credentials);
+    } else if (credentials) {
+      creds = credentials;
+    }
+
+    const serviceKey = creds[AuthKeys.GOOGLE_SERVICE_KEY] ?? {};
+    this.serviceKey =
+      serviceKey && typeof serviceKey === 'string' ? JSON.parse(serviceKey) : serviceKey ?? {};
+    this.client_email = this.serviceKey.client_email;
+    this.private_key = this.serviceKey.private_key;
+    this.project_id = this.serviceKey.project_id;
+    this.access_token = null;
+
+    this.apiKey = creds[AuthKeys.GOOGLE_API_KEY];
+
+    if (options.skipSetOptions) {
+      return;
+    }
    this.setOptions(options);
  }

-  /* Google/PaLM2 specific methods */
+  /* Google specific methods */
  constructUrl() {
-    return `https://us-central1-aiplatform.googleapis.com/v1/projects/${this.project_id}/locations/us-central1/publishers/google/models/${this.modelOptions.model}:predict`;
+    return `${endpointPrefix}/v1/projects/${this.project_id}/locations/${loc}/publishers/${publisher}/models/${this.modelOptions.model}:serverStreamingPredict`;
  }

  async getClient() {
@@ -26,8 +74,7 @@ class GoogleClient extends BaseClient {

    jwtClient.authorize((err) => {
      if (err) {
-        console.error('Error: jwtClient failed to authorize');
-        console.error(err.message);
+        logger.error('jwtClient failed to authorize', err);
        throw err;
      }
    });
@@ -35,6 +82,22 @@ class GoogleClient extends BaseClient {
    return jwtClient;
  }

+  async getAccessToken() {
+    const scopes = ['https://www.googleapis.com/auth/cloud-platform'];
+    const jwtClient = new google.auth.JWT(this.client_email, null, this.private_key, scopes);
+
+    return new Promise((resolve, reject) => {
+      jwtClient.authorize((err, tokens) => {
+        if (err) {
+          logger.error('jwtClient failed to authorize', err);
+          reject(err);
+        } else {
+          resolve(tokens.access_token);
+        }
+      });
+    });
+  }
+
  /* Required Client methods */
  setOptions(options) {
    if (this.options && !this.options.replaceOptions) {
@@ -53,30 +116,35 @@ class GoogleClient extends BaseClient {
      this.options = options;
    }

-    this.options.examples = this.options.examples.filter(
-      (obj) => obj.input.content !== '' && obj.output.content !== '',
-    );
+    this.options.examples = (this.options.examples ?? [])
+      .filter((ex) => ex)
+      .filter((obj) => obj.input.content !== '' && obj.output.content !== '');

-    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 || 'chat-bison',
-      temperature: typeof modelOptions.temperature === 'undefined' ? 0.2 : modelOptions.temperature, // 0 - 1, 0.2 is recommended
-      topP: typeof modelOptions.topP === 'undefined' ? 0.95 : modelOptions.topP, // 0 - 1, default: 0.95
-      topK: typeof modelOptions.topK === 'undefined' ? 40 : modelOptions.topK, // 1-40, default: 40
-      // stop: modelOptions.stop // no stop method for now
-    };
+    this.modelOptions = this.options.modelOptions || {};

-    this.isChatModel = this.modelOptions.model.startsWith('chat-');
+    this.options.attachments?.then((attachments) => this.checkVisionRequest(attachments));
+
+    /** @type {boolean} Whether using a "GenerativeAI" Model */
+    this.isGenerativeModel = this.modelOptions.model.includes('gemini');
+    const { isGenerativeModel } = this;
+    this.isChatModel = !isGenerativeModel && this.modelOptions.model.includes('chat');
    const { isChatModel } = this;
-    this.isTextModel = this.modelOptions.model.startsWith('text-');
+    this.isTextModel =
+      !isGenerativeModel && !isChatModel && /code|text/.test(this.modelOptions.model);
    const { isTextModel } = this;

-    this.maxContextTokens = this.options.maxContextTokens || (isTextModel ? 8000 : 4096);
+    this.maxContextTokens =
+      this.options.maxContextTokens ??
+      getModelMaxTokens(this.modelOptions.model, EModelEndpoint.google);
+
    // 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.maxOutputTokens || 1024;
+    this.maxResponseTokens = this.modelOptions.maxOutputTokens || settings.maxOutputTokens.default;
+
+    if (this.maxContextTokens > 32000) {
+      this.maxContextTokens = this.maxContextTokens - this.maxResponseTokens;
+    }
+
    this.maxPromptTokens =
      this.options.maxPromptTokens || this.maxContextTokens - this.maxResponseTokens;

@@ -88,10 +156,18 @@ class GoogleClient extends BaseClient {
      );
    }

+    this.sender =
+      this.options.sender ??
+      getResponseSender({
+        model: this.modelOptions.model,
+        endpoint: EModelEndpoint.google,
+        modelLabel: this.options.modelLabel,
+      });
+
    this.userLabel = this.options.userLabel || 'User';
    this.modelLabel = this.options.modelLabel || 'Assistant';

-    if (isChatModel) {
+    if (isChatModel || isGenerativeModel) {
      // 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.
@@ -99,8 +175,8 @@ class GoogleClient extends BaseClient {
      this.endToken = '';
      this.gptEncoder = this.constructor.getTokenizer('cl100k_base');
    } else if (isTextModel) {
-      this.startToken = '<|im_start|>';
-      this.endToken = '<|im_end|>';
+      this.startToken = '||>';
+      this.endToken = '';
      this.gptEncoder = this.constructor.getTokenizer('text-davinci-003', true, {
        '<|im_start|>': 100264,
        '<|im_end|>': 100265,
@@ -138,60 +214,363 @@ class GoogleClient extends BaseClient {
    return this;
  }

-  getMessageMapMethod() {
+  /**
+   *
+   * Checks if the model is a vision model based on request attachments and sets the appropriate options:
+   * @param {MongoFile[]} attachments
+   */
+  checkVisionRequest(attachments) {
+    /* Validation vision request */
+    this.defaultVisionModel = this.options.visionModel ?? 'gemini-pro-vision';
+    const availableModels = this.options.modelsConfig?.[EModelEndpoint.google];
+    this.isVisionModel = validateVisionModel({ model: this.modelOptions.model, availableModels });
+
+    if (
+      attachments &&
+      attachments.some((file) => file?.type && file?.type?.includes('image')) &&
+      availableModels?.includes(this.defaultVisionModel) &&
+      !this.isVisionModel
+    ) {
+      this.modelOptions.model = this.defaultVisionModel;
+      this.isVisionModel = true;
+    }
+
+    if (this.isVisionModel && !attachments && this.modelOptions.model.includes('gemini-pro')) {
+      this.modelOptions.model = 'gemini-pro';
+      this.isVisionModel = false;
+    }
+  }
+
+  formatMessages() {
    return ((message) => ({
      author: message?.author ?? (message.isCreatedByUser ? this.userLabel : this.modelLabel),
      content: message?.content ?? message.text,
    })).bind(this);
  }

-  buildMessages(messages = []) {
-    const formattedMessages = messages.map(this.getMessageMapMethod());
+  /**
+   * Formats messages for generative AI
+   * @param {TMessage[]} messages
+   * @returns
+   */
+  async formatGenerativeMessages(messages) {
+    const formattedMessages = [];
+    const attachments = await this.options.attachments;
+    const latestMessage = { ...messages[messages.length - 1] };
+    const files = await this.addImageURLs(latestMessage, attachments, VisionModes.generative);
+    this.options.attachments = files;
+    messages[messages.length - 1] = latestMessage;
+
+    for (const _message of messages) {
+      const role = _message.isCreatedByUser ? this.userLabel : this.modelLabel;
+      const parts = [];
+      parts.push({ text: _message.text });
+      if (!_message.image_urls?.length) {
+        formattedMessages.push({ role, parts });
+        continue;
+      }
+
+      for (const images of _message.image_urls) {
+        if (images.inlineData) {
+          parts.push({ inlineData: images.inlineData });
+        }
+      }
+
+      formattedMessages.push({ role, parts });
+    }
+
+    return formattedMessages;
+  }
+
+  /**
+   *
+   * Adds image URLs to the message object and returns the files
+   *
+   * @param {TMessage[]} messages
+   * @param {MongoFile[]} files
+   * @returns {Promise<MongoFile[]>}
+   */
+  async addImageURLs(message, attachments, mode = '') {
+    const { files, image_urls } = await encodeAndFormat(
+      this.options.req,
+      attachments,
+      EModelEndpoint.google,
+      mode,
+    );
+    message.image_urls = image_urls.length ? image_urls : undefined;
+    return files;
+  }
+
+  /**
+   * Builds the augmented prompt for attachments
+   * TODO: Add File API Support
+   * @param {TMessage[]} messages
+   */
+  async buildAugmentedPrompt(messages = []) {
+    const attachments = await this.options.attachments;
+    const latestMessage = { ...messages[messages.length - 1] };
+    this.contextHandlers = createContextHandlers(this.options.req, latestMessage.text);
+
+    if (this.contextHandlers) {
+      for (const file of attachments) {
+        if (file.embedded) {
+          this.contextHandlers?.processFile(file);
+          continue;
+        }
+      }
+
+      this.augmentedPrompt = await this.contextHandlers.createContext();
+      this.options.promptPrefix = this.augmentedPrompt + this.options.promptPrefix;
+    }
+  }
+
+  async buildVisionMessages(messages = [], parentMessageId) {
+    const attachments = await this.options.attachments;
+    const latestMessage = { ...messages[messages.length - 1] };
+    await this.buildAugmentedPrompt(messages);
+
+    const { prompt } = await this.buildMessagesPrompt(messages, parentMessageId);
+
+    const files = await this.addImageURLs(latestMessage, attachments);
+
+    this.options.attachments = files;
+
+    latestMessage.text = prompt;
+
+    const payload = {
+      instances: [
+        {
+          messages: [new HumanMessage(formatMessage({ message: latestMessage }))],
+        },
+      ],
+      parameters: this.modelOptions,
+    };
+    return { prompt: payload };
+  }
+
+  /** @param {TMessage[]} [messages=[]]  */
+  async buildGenerativeMessages(messages = []) {
+    this.userLabel = 'user';
+    this.modelLabel = 'model';
+    const promises = [];
+    promises.push(await this.formatGenerativeMessages(messages));
+    promises.push(this.buildAugmentedPrompt(messages));
+    const [formattedMessages] = await Promise.all(promises);
+    return { prompt: formattedMessages };
+  }
+
+  async buildMessages(messages = [], parentMessageId) {
+    if (!this.isGenerativeModel && !this.project_id) {
+      throw new Error(
+        '[GoogleClient] a Service Account JSON Key is required for PaLM 2 and Codey models (Vertex AI)',
+      );
+    }
+
+    if (!this.project_id && this.modelOptions.model.includes('1.5')) {
+      return await this.buildGenerativeMessages(messages);
+    }
+
+    if (this.options.attachments && this.isGenerativeModel) {
+      return this.buildVisionMessages(messages, parentMessageId);
+    }
+
+    if (this.isTextModel) {
+      return this.buildMessagesPrompt(messages, parentMessageId);
+    }
+
    let payload = {
      instances: [
        {
-          messages: formattedMessages,
+          messages: messages
+            .map(this.formatMessages())
+            .map((msg) => ({ ...msg, role: msg.author === 'User' ? 'user' : 'assistant' }))
+            .map((message) => formatMessage({ message, langChain: true })),
        },
      ],
-      parameters: this.options.modelOptions,
+      parameters: this.modelOptions,
    };

-    if (this.options.promptPrefix) {
-      payload.instances[0].context = this.options.promptPrefix;
+    let promptPrefix = (this.options.promptPrefix ?? '').trim();
+    if (typeof this.options.artifactsPrompt === 'string' && this.options.artifactsPrompt) {
+      promptPrefix = `${promptPrefix ?? ''}\n${this.options.artifactsPrompt}`.trim();
+    }
+
+    if (promptPrefix) {
+      payload.instances[0].context = promptPrefix;
    }

    if (this.options.examples.length > 0) {
      payload.instances[0].examples = this.options.examples;
    }

-    /* TO-DO: text model needs more context since it can't process an array of messages */
-    if (this.isTextModel) {
-      payload.instances = [
-        {
-          prompt: messages[messages.length - 1].content,
-        },
-      ];
-    }
-
-    if (this.options.debug) {
-      console.debug('GoogleClient buildMessages');
-      console.dir(payload, { depth: null });
-    }
+    logger.debug('[GoogleClient] buildMessages', payload);

    return { prompt: payload };
  }

-  async getCompletion(payload, abortController = null) {
+  async buildMessagesPrompt(messages, parentMessageId) {
+    const orderedMessages = this.constructor.getMessagesForConversation({
+      messages,
+      parentMessageId,
+    });
+
+    logger.debug('[GoogleClient]', {
+      orderedMessages,
+      parentMessageId,
+    });
+
+    const formattedMessages = orderedMessages.map((message) => ({
+      author: message.isCreatedByUser ? this.userLabel : this.modelLabel,
+      content: message?.content ?? message.text,
+    }));
+
+    let lastAuthor = '';
+    let groupedMessages = [];
+
+    for (let message of formattedMessages) {
+      // If last author is not same as current author, add to new group
+      if (lastAuthor !== message.author) {
+        groupedMessages.push({
+          author: message.author,
+          content: [message.content],
+        });
+        lastAuthor = message.author;
+        // If same author, append content to the last group
+      } else {
+        groupedMessages[groupedMessages.length - 1].content.push(message.content);
+      }
+    }
+
+    let identityPrefix = '';
+    if (this.options.userLabel) {
+      identityPrefix = `\nHuman's name: ${this.options.userLabel}`;
+    }
+
+    if (this.options.modelLabel) {
+      identityPrefix = `${identityPrefix}\nYou are ${this.options.modelLabel}`;
+    }
+
+    let promptPrefix = (this.options.promptPrefix ?? '').trim();
+    if (typeof this.options.artifactsPrompt === 'string' && this.options.artifactsPrompt) {
+      promptPrefix = `${promptPrefix ?? ''}\n${this.options.artifactsPrompt}`.trim();
+    }
+    if (promptPrefix) {
+      // If the prompt prefix doesn't end with the end token, add it.
+      if (!promptPrefix.endsWith(`${this.endToken}`)) {
+        promptPrefix = `${promptPrefix.trim()}${this.endToken}\n\n`;
+      }
+      promptPrefix = `\nContext:\n${promptPrefix}`;
+    }
+
+    if (identityPrefix) {
+      promptPrefix = `${identityPrefix}${promptPrefix}`;
+    }
+
+    // Prompt AI to respond, empty if last message was from AI
+    let isEdited = lastAuthor === this.modelLabel;
+    const promptSuffix = isEdited ? '' : `${promptPrefix}\n\n${this.modelLabel}:\n`;
+    let currentTokenCount = isEdited
+      ? this.getTokenCount(promptPrefix)
+      : this.getTokenCount(promptSuffix);
+
+    let promptBody = '';
+    const maxTokenCount = this.maxPromptTokens;
+
+    const context = [];
+
+    // Iterate backwards through the messages, adding them to the prompt until we reach the max token count.
+    // Do this within a recursive async function so that it doesn't block the event loop for too long.
+    // Also, remove the next message when the message that puts us over the token limit is created by the user.
+    // Otherwise, remove only the exceeding message. This is due to Anthropic's strict payload rule to start with "Human:".
+    const nextMessage = {
+      remove: false,
+      tokenCount: 0,
+      messageString: '',
+    };
+
+    const buildPromptBody = async () => {
+      if (currentTokenCount < maxTokenCount && groupedMessages.length > 0) {
+        const message = groupedMessages.pop();
+        const isCreatedByUser = message.author === this.userLabel;
+        // Use promptPrefix if message is edited assistant'
+        const messagePrefix =
+          isCreatedByUser || !isEdited
+            ? `\n\n${message.author}:`
+            : `${promptPrefix}\n\n${message.author}:`;
+        const messageString = `${messagePrefix}\n${message.content}${this.endToken}\n`;
+        let newPromptBody = `${messageString}${promptBody}`;
+
+        context.unshift(message);
+
+        const tokenCountForMessage = this.getTokenCount(messageString);
+        const newTokenCount = currentTokenCount + tokenCountForMessage;
+
+        if (!isCreatedByUser) {
+          nextMessage.messageString = messageString;
+          nextMessage.tokenCount = tokenCountForMessage;
+        }
+
+        if (newTokenCount > maxTokenCount) {
+          if (!promptBody) {
+            // 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 ${newTokenCount} tokens long.`,
+            );
+          }
+
+          // Otherwise, ths message would put us over the token limit, so don't add it.
+          // if created by user, remove next message, otherwise remove only this message
+          if (isCreatedByUser) {
+            nextMessage.remove = true;
+          }
+
+          return false;
+        }
+        promptBody = newPromptBody;
+        currentTokenCount = newTokenCount;
+
+        // Switch off isEdited after using it for the first time
+        if (isEdited) {
+          isEdited = false;
+        }
+
+        // wait for next tick to avoid blocking the event loop
+        await new Promise((resolve) => setImmediate(resolve));
+        return buildPromptBody();
+      }
+      return true;
+    };
+
+    await buildPromptBody();
+
+    if (nextMessage.remove) {
+      promptBody = promptBody.replace(nextMessage.messageString, '');
+      currentTokenCount -= nextMessage.tokenCount;
+      context.shift();
+    }
+
+    let prompt = `${promptBody}${promptSuffix}`.trim();
+
+    // Add 2 tokens for metadata after all messages have been counted.
+    currentTokenCount += 2;
+
+    // Use up to `this.maxContextTokens` tokens (prompt + response), but try to leave `this.maxTokens` tokens for the response.
+    this.modelOptions.maxOutputTokens = Math.min(
+      this.maxContextTokens - currentTokenCount,
+      this.maxResponseTokens,
+    );
+
+    return { prompt, context };
+  }
+
+  async _getCompletion(payload, abortController = null) {
    if (!abortController) {
      abortController = new AbortController();
    }
    const { debug } = this.options;
    const url = this.completionsUrl;
    if (debug) {
-      console.debug();
-      console.debug(url);
-      console.debug(this.modelOptions);
-      console.debug();
+      logger.debug('GoogleClient _getCompletion', { url, payload });
    }
    const opts = {
      method: 'POST',
@@ -208,54 +587,324 @@ class GoogleClient extends BaseClient {

    const client = await this.getClient();
    const res = await client.request({ url, method: 'POST', data: payload });
-    console.dir(res.data, { depth: null });
+    logger.debug('GoogleClient _getCompletion', { res });
    return res.data;
  }

+  createLLM(clientOptions) {
+    const model = clientOptions.modelName ?? clientOptions.model;
+    if (this.project_id && this.isTextModel) {
+      logger.debug('Creating Google VertexAI client');
+      return new GoogleVertexAI(clientOptions);
+    } else if (this.project_id && this.isChatModel) {
+      logger.debug('Creating Chat Google VertexAI client');
+      return new ChatGoogleVertexAI(clientOptions);
+    } else if (this.project_id) {
+      logger.debug('Creating VertexAI client');
+      return new ChatVertexAI(clientOptions);
+    } else if (model.includes('1.5')) {
+      logger.debug('Creating GenAI client');
+      return new GenAI(this.apiKey).getGenerativeModel(
+        {
+          ...clientOptions,
+          model,
+        },
+        { apiVersion: 'v1beta' },
+      );
+    }
+
+    logger.debug('Creating Chat Google Generative AI client');
+    return new ChatGoogleGenerativeAI({ ...clientOptions, apiKey: this.apiKey });
+  }
+
+  async getCompletion(_payload, options = {}) {
+    const { parameters, instances } = _payload;
+    const { onProgress, abortController } = options;
+    const streamRate = this.options.streamRate ?? Constants.DEFAULT_STREAM_RATE;
+    const { messages: _messages, context, examples: _examples } = instances?.[0] ?? {};
+
+    let examples;
+
+    let clientOptions = { ...parameters, maxRetries: 2 };
+
+    if (this.project_id) {
+      clientOptions['authOptions'] = {
+        credentials: {
+          ...this.serviceKey,
+        },
+        projectId: this.project_id,
+      };
+    }
+
+    if (!parameters) {
+      clientOptions = { ...clientOptions, ...this.modelOptions };
+    }
+
+    if (this.isGenerativeModel && !this.project_id) {
+      clientOptions.modelName = clientOptions.model;
+      delete clientOptions.model;
+    }
+
+    if (_examples && _examples.length) {
+      examples = _examples
+        .map((ex) => {
+          const { input, output } = ex;
+          if (!input || !output) {
+            return undefined;
+          }
+          return {
+            input: new HumanMessage(input.content),
+            output: new AIMessage(output.content),
+          };
+        })
+        .filter((ex) => ex);
+
+      clientOptions.examples = examples;
+    }
+
+    const model = this.createLLM(clientOptions);
+
+    let reply = '';
+    const messages = this.isTextModel ? _payload.trim() : _messages;
+
+    if (!this.isVisionModel && context && messages?.length > 0) {
+      messages.unshift(new SystemMessage(context));
+    }
+
+    const modelName = clientOptions.modelName ?? clientOptions.model ?? '';
+    if (modelName?.includes('1.5') && !this.project_id) {
+      const client = model;
+      const requestOptions = {
+        contents: _payload,
+      };
+
+      let promptPrefix = (this.options.promptPrefix ?? '').trim();
+      if (typeof this.options.artifactsPrompt === 'string' && this.options.artifactsPrompt) {
+        promptPrefix = `${promptPrefix ?? ''}\n${this.options.artifactsPrompt}`.trim();
+      }
+
+      if (this.options?.promptPrefix?.length) {
+        requestOptions.systemInstruction = {
+          parts: [
+            {
+              text: promptPrefix,
+            },
+          ],
+        };
+      }
+
+      requestOptions.safetySettings = _payload.safetySettings;
+
+      const delay = modelName.includes('flash') ? 8 : 14;
+      const result = await client.generateContentStream(requestOptions);
+      for await (const chunk of result.stream) {
+        const chunkText = chunk.text();
+        await this.generateTextStream(chunkText, onProgress, {
+          delay,
+        });
+        reply += chunkText;
+        await sleep(streamRate);
+      }
+      return reply;
+    }
+
+    const stream = await model.stream(messages, {
+      signal: abortController.signal,
+      timeout: 7000,
+      safetySettings: _payload.safetySettings,
+    });
+
+    let delay = this.options.streamRate || 8;
+
+    if (!this.options.streamRate) {
+      if (this.isGenerativeModel) {
+        delay = 12;
+      }
+      if (modelName.includes('flash')) {
+        delay = 5;
+      }
+    }
+
+    for await (const chunk of stream) {
+      const chunkText = chunk?.content ?? chunk;
+      await this.generateTextStream(chunkText, onProgress, {
+        delay,
+      });
+      reply += chunkText;
+    }
+
+    return reply;
+  }
+
+  /**
+   * Stripped-down logic for generating a title. This uses the non-streaming APIs, since the user does not see titles streaming
+   */
+  async titleChatCompletion(_payload, options = {}) {
+    const { abortController } = options;
+    const { parameters, instances } = _payload;
+    const { messages: _messages, examples: _examples } = instances?.[0] ?? {};
+
+    let clientOptions = { ...parameters, maxRetries: 2 };
+
+    logger.debug('Initialized title client options');
+
+    if (this.project_id) {
+      clientOptions['authOptions'] = {
+        credentials: {
+          ...this.serviceKey,
+        },
+        projectId: this.project_id,
+      };
+    }
+
+    if (!parameters) {
+      clientOptions = { ...clientOptions, ...this.modelOptions };
+    }
+
+    if (this.isGenerativeModel && !this.project_id) {
+      clientOptions.modelName = clientOptions.model;
+      delete clientOptions.model;
+    }
+
+    const model = this.createLLM(clientOptions);
+
+    let reply = '';
+    const messages = this.isTextModel ? _payload.trim() : _messages;
+
+    const modelName = clientOptions.modelName ?? clientOptions.model ?? '';
+    if (modelName?.includes('1.5') && !this.project_id) {
+      logger.debug('Identified titling model as 1.5 version');
+      /** @type {GenerativeModel} */
+      const client = model;
+      const requestOptions = {
+        contents: _payload,
+      };
+
+      let promptPrefix = (this.options.promptPrefix ?? '').trim();
+      if (typeof this.options.artifactsPrompt === 'string' && this.options.artifactsPrompt) {
+        promptPrefix = `${promptPrefix ?? ''}\n${this.options.artifactsPrompt}`.trim();
+      }
+
+      if (this.options?.promptPrefix?.length) {
+        requestOptions.systemInstruction = {
+          parts: [
+            {
+              text: promptPrefix,
+            },
+          ],
+        };
+      }
+
+      const safetySettings = _payload.safetySettings;
+      requestOptions.safetySettings = safetySettings;
+
+      const result = await client.generateContent(requestOptions);
+
+      reply = result.response?.text();
+
+      return reply;
+    } else {
+      logger.debug('Beginning titling');
+      const safetySettings = _payload.safetySettings;
+
+      const titleResponse = await model.invoke(messages, {
+        signal: abortController.signal,
+        timeout: 7000,
+        safetySettings: safetySettings,
+      });
+
+      reply = titleResponse.content;
+      // TODO: RECORD TOKEN USAGE
+      return reply;
+    }
+  }
+
+  async titleConvo({ text, responseText = '' }) {
+    let title = 'New Chat';
+    const convo = `||>User:
+"${truncateText(text)}"
+||>Response:
+"${JSON.stringify(truncateText(responseText))}"`;
+
+    let { prompt: payload } = await this.buildMessages([
+      {
+        text: `Please generate ${titleInstruction}
+
+    ${convo}
+    
+    ||>Title:`,
+        isCreatedByUser: true,
+        author: this.userLabel,
+      },
+    ]);
+
+    if (this.isVisionModel) {
+      logger.warn(
+        `Current vision model does not support titling without an attachment; falling back to default model ${settings.model.default}`,
+      );
+
+      payload.parameters = { ...payload.parameters, model: settings.model.default };
+    }
+
+    try {
+      title = await this.titleChatCompletion(payload, {
+        abortController: new AbortController(),
+        onProgress: () => {},
+      });
+    } catch (e) {
+      logger.error('[GoogleClient] There was an issue generating the title', e);
+    }
+    logger.debug(`Title response: ${title}`);
+    return title;
+  }
+
  getSaveOptions() {
    return {
+      artifacts: this.options.artifacts,
      promptPrefix: this.options.promptPrefix,
      modelLabel: this.options.modelLabel,
+      iconURL: this.options.iconURL,
+      greeting: this.options.greeting,
+      spec: this.options.spec,
      ...this.modelOptions,
    };
  }

  getBuildMessagesOptions() {
-    // console.log('GoogleClient doesn\'t use getBuildMessagesOptions');
+    // logger.debug('GoogleClient doesn\'t use getBuildMessagesOptions');
  }

  async sendCompletion(payload, opts = {}) {
-    console.log('GoogleClient: sendcompletion', payload, opts);
+    payload.safetySettings = this.getSafetySettings();
+
    let reply = '';
-    let blocked = false;
-    try {
-      const result = await this.getCompletion(payload, opts.abortController);
-      blocked = result?.predictions?.[0]?.safetyAttributes?.blocked;
-      reply =
-        result?.predictions?.[0]?.candidates?.[0]?.content ||
-        result?.predictions?.[0]?.content ||
-        '';
-      if (blocked === true) {
-        reply = `Google blocked a proper response to your message:\n${JSON.stringify(
-          result.predictions[0].safetyAttributes,
-        )}${reply.length > 0 ? `\nAI Response:\n${reply}` : ''}`;
-      }
-      if (this.options.debug) {
-        console.debug('result');
-        console.debug(result);
-      }
-    } catch (err) {
-      console.error('Error: failed to send completion to Google');
-      console.error(err.message);
-    }
-
-    if (!blocked) {
-      await this.generateTextStream(reply, opts.onProgress, { delay: 0.5 });
-    }
-
+    reply = await this.getCompletion(payload, opts);
    return reply.trim();
  }

+  getSafetySettings() {
+    return [
+      {
+        category: 'HARM_CATEGORY_SEXUALLY_EXPLICIT',
+        threshold:
+          process.env.GOOGLE_SAFETY_SEXUALLY_EXPLICIT || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
+      },
+      {
+        category: 'HARM_CATEGORY_HATE_SPEECH',
+        threshold: process.env.GOOGLE_SAFETY_HATE_SPEECH || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
+      },
+      {
+        category: 'HARM_CATEGORY_HARASSMENT',
+        threshold: process.env.GOOGLE_SAFETY_HARASSMENT || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
+      },
+      {
+        category: 'HARM_CATEGORY_DANGEROUS_CONTENT',
+        threshold:
+          process.env.GOOGLE_SAFETY_DANGEROUS_CONTENT || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
+      },
+    ];
+  }
+
  /* TO-DO: Handle tokens with Google tokenization NOTE: these are required */
  static getTokenizer(encoding, isModelName = false, extendSpecialTokens = {}) {
    if (tokenizersCache[encoding]) {
--- a/api/app/clients/OllamaClient.js
+++ b/api/app/clients/OllamaClient.js
@@ -0,0 +1,159 @@
+const { z } = require('zod');
+const axios = require('axios');
+const { Ollama } = require('ollama');
+const { Constants } = require('librechat-data-provider');
+const { deriveBaseURL } = require('~/utils');
+const { sleep } = require('~/server/utils');
+const { logger } = require('~/config');
+
+const ollamaPayloadSchema = z.object({
+  mirostat: z.number().optional(),
+  mirostat_eta: z.number().optional(),
+  mirostat_tau: z.number().optional(),
+  num_ctx: z.number().optional(),
+  repeat_last_n: z.number().optional(),
+  repeat_penalty: z.number().optional(),
+  temperature: z.number().optional(),
+  seed: z.number().nullable().optional(),
+  stop: z.array(z.string()).optional(),
+  tfs_z: z.number().optional(),
+  num_predict: z.number().optional(),
+  top_k: z.number().optional(),
+  top_p: z.number().optional(),
+  stream: z.optional(z.boolean()),
+  model: z.string(),
+});
+
+/**
+ * @param {string} imageUrl
+ * @returns {string}
+ * @throws {Error}
+ */
+const getValidBase64 = (imageUrl) => {
+  const parts = imageUrl.split(';base64,');
+
+  if (parts.length === 2) {
+    return parts[1];
+  } else {
+    logger.error('Invalid or no Base64 string found in URL.');
+  }
+};
+
+class OllamaClient {
+  constructor(options = {}) {
+    const host = deriveBaseURL(options.baseURL ?? 'http://localhost:11434');
+    this.streamRate = options.streamRate ?? Constants.DEFAULT_STREAM_RATE;
+    /** @type {Ollama} */
+    this.client = new Ollama({ host });
+  }
+
+  /**
+   * Fetches Ollama models from the specified base API path.
+   * @param {string} baseURL
+   * @returns {Promise<string[]>} The Ollama models.
+   */
+  static async fetchModels(baseURL) {
+    let models = [];
+    if (!baseURL) {
+      return models;
+    }
+    try {
+      const ollamaEndpoint = deriveBaseURL(baseURL);
+      /** @type {Promise<AxiosResponse<OllamaListResponse>>} */
+      const response = await axios.get(`${ollamaEndpoint}/api/tags`);
+      models = response.data.models.map((tag) => tag.name);
+      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).';
+      logger.error(logMessage, error);
+      return [];
+    }
+  }
+
+  /**
+   * @param {ChatCompletionMessage[]} messages
+   * @returns {OllamaMessage[]}
+   */
+  static formatOpenAIMessages(messages) {
+    const ollamaMessages = [];
+
+    for (const message of messages) {
+      if (typeof message.content === 'string') {
+        ollamaMessages.push({
+          role: message.role,
+          content: message.content,
+        });
+        continue;
+      }
+
+      let aggregatedText = '';
+      let imageUrls = [];
+
+      for (const content of message.content) {
+        if (content.type === 'text') {
+          aggregatedText += content.text + ' ';
+        } else if (content.type === 'image_url') {
+          imageUrls.push(getValidBase64(content.image_url.url));
+        }
+      }
+
+      const ollamaMessage = {
+        role: message.role,
+        content: aggregatedText.trim(),
+      };
+
+      if (imageUrls.length > 0) {
+        ollamaMessage.images = imageUrls;
+      }
+
+      ollamaMessages.push(ollamaMessage);
+    }
+
+    return ollamaMessages;
+  }
+
+  /***
+   * @param {Object} params
+   * @param {ChatCompletionPayload} params.payload
+   * @param {onTokenProgress} params.onProgress
+   * @param {AbortController} params.abortController
+   */
+  async chatCompletion({ payload, onProgress, abortController = null }) {
+    let intermediateReply = '';
+
+    const parameters = ollamaPayloadSchema.parse(payload);
+    const messages = OllamaClient.formatOpenAIMessages(payload.messages);
+
+    if (parameters.stream) {
+      const stream = await this.client.chat({
+        messages,
+        ...parameters,
+      });
+
+      for await (const chunk of stream) {
+        const token = chunk.message.content;
+        intermediateReply += token;
+        onProgress(token);
+        if (abortController.signal.aborted) {
+          stream.controller.abort();
+          break;
+        }
+
+        await sleep(this.streamRate);
+      }
+    }
+    // TODO: regular completion
+    else {
+      // const generation = await this.client.generate(payload);
+    }
+
+    return intermediateReply;
+  }
+  catch(err) {
+    logger.error('[OllamaClient.chatCompletion]', err);
+    throw err;
+  }
+}
+
+module.exports = { OllamaClient, ollamaPayloadSchema };
--- a/api/app/clients/OpenAIClient.js
+++ b/api/app/clients/OpenAIClient.js
--- a/api/app/clients/PluginsClient.js
+++ b/api/app/clients/PluginsClient.js
@@ -1,11 +1,19 @@
 const OpenAIClient = require('./OpenAIClient');
 const { CallbackManager } = require('langchain/callbacks');
-const { HumanChatMessage, AIChatMessage } = require('langchain/schema');
+const { CacheKeys, Time } = require('librechat-data-provider');
+const { BufferMemory, ChatMessageHistory } = require('langchain/memory');
 const { initializeCustomAgent, initializeFunctionsAgent } = require('./agents');
 const { addImages, buildErrorInput, buildPromptPrefix } = require('./output_parsers');
+const { processFileURL } = require('~/server/services/Files/process');
+const { EModelEndpoint } = require('librechat-data-provider');
+const { formatLangChainMessages } = require('./prompts');
+const checkBalance = require('~/models/checkBalance');
 const { SelfReflectionTool } = require('./tools');
+const { isEnabled } = require('~/server/utils');
+const { extractBaseURL } = require('~/utils');
 const { loadTools } = require('./tools/util');
-const { createLLM } = require('./llm');
+const { getLogStores } = require('~/cache');
+const { logger } = require('~/config');

 class PluginsClient extends OpenAIClient {
  constructor(apiKey, options = {}) {
@@ -25,23 +33,24 @@ class PluginsClient extends OpenAIClient {

    super.setOptions(options);

-    if (this.functionsAgent && this.agentOptions.model && !this.useOpenRouter) {
-      this.agentOptions.model = this.getFunctionModelName(this.agentOptions.model);
-    }
-
    this.isGpt3 = this.modelOptions?.model?.includes('gpt-3');

    if (this.options.reverseProxyUrl) {
-      this.langchainProxy = this.options.reverseProxyUrl.match(/.*v1/)[0];
+      this.langchainProxy = extractBaseURL(this.options.reverseProxyUrl);
    }
  }

  getSaveOptions() {
    return {
+      artifacts: this.options.artifacts,
      chatGptLabel: this.options.chatGptLabel,
      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,
    };
  }

@@ -50,9 +59,11 @@ class PluginsClient extends OpenAIClient {
  }

  getFunctionModelName(input) {
-    if (input.startsWith('gpt-3.5-turbo')) {
+    if (/-(?!0314)\d{4}/.test(input)) {
+      return input;
+    } else if (input.includes('gpt-3.5-turbo')) {
      return 'gpt-3.5-turbo';
-    } else if (input.startsWith('gpt-4')) {
+    } else if (input.includes('gpt-4')) {
      return 'gpt-4';
    } else {
      return 'gpt-3.5-turbo';
@@ -73,34 +84,27 @@ class PluginsClient extends OpenAIClient {
      temperature: this.agentOptions.temperature,
    };

-    const configOptions = {};
-
-    if (this.langchainProxy) {
-      configOptions.basePath = this.langchainProxy;
-    }
-
-    if (this.useOpenRouter) {
-      configOptions.basePath = 'https://openrouter.ai/api/v1';
-      configOptions.baseOptions = {
-        headers: {
-          'HTTP-Referer': 'https://librechat.ai',
-          'X-Title': 'LibreChat',
-        },
-      };
-    }
-
-    const model = createLLM({
-      modelOptions,
-      configOptions,
-      openAIApiKey: this.openAIApiKey,
-      azure: this.azure,
+    const model = this.initializeLLM({
+      ...modelOptions,
+      context: 'plugins',
+      initialMessageCount: this.currentMessages.length + 1,
    });

-    if (this.options.debug) {
-      console.debug(
-        `<-----Agent Model: ${model.modelName} | Temp: ${model.temperature} | Functions: ${this.functionsAgent}----->`,
-      );
-    }
+    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),
+    });

    this.tools = await loadTools({
      user,
@@ -108,9 +112,12 @@ class PluginsClient extends OpenAIClient {
      tools: this.options.tools,
      functions: this.functionsAgent,
      options: {
+        memory,
+        signal: this.abortController.signal,
        openAIApiKey: this.openAIApiKey,
        conversationId: this.conversationId,
-        debug: this.options?.debug,
+        fileStrategy: this.options.req.app.locals.fileStrategy,
+        processFileURL,
        message,
      },
    });
@@ -121,44 +128,40 @@ class PluginsClient extends OpenAIClient {
      return;
    }

-    if (this.options.debug) {
-      console.debug('Requested Tools');
-      console.debug(this.options.tools);
-      console.debug('Loaded Tools');
-      console.debug(this.tools.map((tool) => tool.name));
-    }
+    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);

-      if (this.options.debug) {
-        console.debug('Latest Agent Action ', this.actions[this.actions.length - 1]);
-      }
+      logger.debug('[PluginsClient] Latest Agent Action ', this.actions[this.actions.length - 1]);

      if (typeof callback === 'function') {
        callback(action, runId);
      }
    };

-    // Map Messages to Langchain format
-    const pastMessages = this.currentMessages
-      .slice(0, -1)
-      .map((msg) =>
-        msg?.isCreatedByUser || msg?.role?.toLowerCase() === 'user'
-          ? new HumanChatMessage(msg.text)
-          : new AIChatMessage(msg.text),
-      );
-
    // 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,
-      currentDateString: this.currentDateString,
+      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);
@@ -171,9 +174,7 @@ class PluginsClient extends OpenAIClient {
      }),
    });

-    if (this.options.debug) {
-      console.debug('Loaded agent.');
-    }
+    logger.debug('[PluginsClient] Loaded agent.');
  }

  async executorCall(message, { signal, stream, onToolStart, onToolEnd }) {
@@ -189,12 +190,10 @@ class PluginsClient extends OpenAIClient {
      });
      const input = attempts > 1 ? errorInput : message;

-      if (this.options.debug) {
-        console.debug(`Attempt ${attempts} of ${maxAttempts}`);
-      }
+      logger.debug(`[PluginsClient] Attempt ${attempts} of ${maxAttempts}`);

-      if (this.options.debug && errorMessage.length > 0) {
-        console.debug('Caught error, input:', input);
+      if (errorMessage.length > 0) {
+        logger.debug('[PluginsClient] Caught error, input: ' + JSON.stringify(input));
      }

      try {
@@ -217,41 +216,78 @@ class PluginsClient extends OpenAIClient {
        ]);
        break; // Exit the loop if the function call is successful
      } catch (err) {
-        console.error(err);
-        errorMessage = err.message;
-        let content = '';
-        if (content) {
-          errorMessage = content;
-          break;
-        }
+        logger.error('[PluginsClient] executorCall error:', err);
        if (attempts === maxAttempts) {
-          this.result.output = `Encountered an error while attempting to respond. Error: ${err.message}`;
+          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;
-          this.result.errorMessage = errorMessage;
          break;
        }
      }
    }
  }

+  /**
+   *
+   * @param {TMessage} responseMessage
+   * @param {Partial<TMessage>} saveOptions
+   * @param {string} user
+   * @returns
+   */
  async handleResponseMessage(responseMessage, saveOptions, user) {
-    responseMessage.tokenCount = this.getTokenCountForResponse(responseMessage);
-    responseMessage.completionTokens = responseMessage.tokenCount;
-    await this.saveMessageToDatabase(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);
+    const messageCache = getLogStores(CacheKeys.MESSAGES);
+    messageCache.set(
+      responseMessage.messageId,
+      {
+        text: responseMessage.text,
+        complete: true,
+      },
+      Time.FIVE_MINUTES,
+    );
    delete responseMessage.tokenCount;
-    return { ...responseMessage, ...this.result };
+    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);
    }
-    if (this.options.debug) {
-      console.log('Plugins sendMessage', message, opts);
-    }
+
+    logger.debug('[PluginsClient] sendMessage', { userMessageText: message, opts });
    const {
      user,
      isEdited,
@@ -265,14 +301,20 @@ class PluginsClient extends OpenAIClient {
      onToolEnd,
    } = await this.handleStartMethods(message, opts);

-    this.conversationId = conversationId;
+    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,
-      messages,
    } = await this.buildMessages(
      this.currentMessages,
      userMessage.messageId,
@@ -283,25 +325,46 @@ class PluginsClient extends OpenAIClient {
    );

    if (tokenCountMap) {
-      console.dir(tokenCountMap, { depth: null });
+      logger.debug('[PluginsClient] tokenCountMap', { tokenCountMap });
      if (tokenCountMap[userMessage.messageId]) {
        userMessage.tokenCount = tokenCountMap[userMessage.messageId];
-        console.log('userMessage.tokenCount', userMessage.tokenCount);
+        logger.debug('[PluginsClient] userMessage.tokenCount', userMessage.tokenCount);
      }
-      payload = payload.map((message) => {
-        const messageWithoutTokenCount = message;
-        delete messageWithoutTokenCount.tokenCount;
-        return messageWithoutTokenCount;
-      });
      this.handleTokenCountMap(tokenCountMap);
    }

    this.result = {};
-    if (messages) {
-      this.currentMessages = messages;
+    if (payload) {
+      this.currentMessages = payload;
    }
-    await this.saveMessageToDatabase(userMessage, saveOptions, user);
+
+    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,
@@ -337,11 +400,19 @@ class PluginsClient extends OpenAIClient {
      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);
    }
@@ -353,10 +424,7 @@ class PluginsClient extends OpenAIClient {
      return await this.handleResponseMessage(responseMessage, saveOptions, user);
    }

-    if (this.options.debug) {
-      console.debug('Plugins completion phase: this.result');
-      console.debug(this.result);
-    }
+    logger.debug('[PluginsClient] Completion phase: this.result', this.result);

    const promptPrefix = buildPromptPrefix({
      result: this.result,
@@ -364,28 +432,20 @@ class PluginsClient extends OpenAIClient {
      functionsAgent: this.functionsAgent,
    });

-    if (this.options.debug) {
-      console.debug('Plugins: promptPrefix');
-      console.debug(promptPrefix);
-    }
+    logger.debug('[PluginsClient]', { promptPrefix });

    payload = await this.buildCompletionPrompt({
      messages: this.currentMessages,
      promptPrefix,
    });

-    if (this.options.debug) {
-      console.debug('buildCompletionPrompt Payload');
-      console.debug(payload);
-    }
+    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 }) {
-    if (this.options.debug) {
-      console.debug('buildCompletionPrompt messages', messages);
-    }
+    logger.debug('[PluginsClient] buildCompletionPrompt messages', messages);

    const orderedMessages = messages;
    let promptPrefix = _promptPrefix.trim();
@@ -431,7 +491,9 @@ class PluginsClient extends OpenAIClient {
        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${message.text}${this.endToken}\n`;
+        let messageString = `${this.startToken}${roleLabel}:\n${
+          message.text ?? message.content ?? ''
+        }${this.endToken}\n`;
        let newPromptBody = `${messageString}${promptBody}`;

        const tokenCountForMessage = this.getTokenCount(messageString);
--- a/api/app/clients/TextStream.js
+++ b/api/app/clients/TextStream.js
@@ -1,4 +1,5 @@
 const { Readable } = require('stream');
+const { logger } = require('~/config');

 class TextStream extends Readable {
  constructor(text, options = {}) {
@@ -38,7 +39,7 @@ class TextStream extends Readable {
      });

      this.on('end', () => {
-        // console.log('Stream ended');
+        // logger.debug('[processTextStream] Stream ended');
        resolve();
      });

@@ -50,7 +51,7 @@ class TextStream extends Readable {
    try {
      await streamPromise;
    } catch (err) {
-      console.error('Error processing text stream:', err);
+      logger.error('[processTextStream] Error in text stream:', err);
      // Handle the error appropriately, e.g., return an error message or throw an error
    }
  }
--- a/api/app/clients/agents/CustomAgent/CustomAgent.js
+++ b/api/app/clients/agents/CustomAgent/CustomAgent.js
@@ -16,11 +16,11 @@ class CustomAgent extends ZeroShotAgent {
    const inputVariables = ['input', 'chat_history', 'agent_scratchpad'];

    let prefix, instructions, suffix;
-    if (model.startsWith('gpt-3')) {
+    if (model.includes('gpt-3')) {
      prefix = gpt3.prefix;
      instructions = gpt3.instructions;
      suffix = gpt3.suffix;
-    } else if (model.startsWith('gpt-4')) {
+    } else if (model.includes('gpt-4')) {
      prefix = gpt4.prefix;
      instructions = gpt4.instructions;
      suffix = gpt4.suffix;
--- a/api/app/clients/agents/CustomAgent/initializeCustomAgent.js
+++ b/api/app/clients/agents/CustomAgent/initializeCustomAgent.js
@@ -13,12 +13,20 @@ const initializeCustomAgent = async ({
  tools,
  model,
  pastMessages,
+  customName,
+  customInstructions,
  currentDateString,
  ...rest
 }) => {
  let prompt = CustomAgent.createPrompt(tools, { currentDateString, model: model.modelName });
+  if (customName) {
+    prompt = `You are "${customName}".\n${prompt}`;
+  }
+  if (customInstructions) {
+    prompt = `${prompt}\n${customInstructions}`;
+  }

-  const chatPrompt = ChatPromptTemplate.fromPromptMessages([
+  const chatPrompt = ChatPromptTemplate.fromMessages([
    new SystemMessagePromptTemplate(prompt),
    HumanMessagePromptTemplate.fromTemplate(`{chat_history}
 Query: {input}
@@ -28,6 +36,7 @@ Query: {input}
  const outputParser = new CustomOutputParser({ tools });

  const memory = new BufferMemory({
+    llm: model,
    chatHistory: new ChatMessageHistory(pastMessages),
    // returnMessages: true, // commenting this out retains memory
    memoryKey: 'chat_history',
--- a/api/app/clients/agents/CustomAgent/instructions.js
+++ b/api/app/clients/agents/CustomAgent/instructions.js
@@ -1,44 +1,3 @@
-/*
-module.exports = `You are ChatGPT, a Large Language model with useful tools.
-
-Talk to the human and provide meaningful answers when questions are asked.
-
-Use the tools when you need them, but use your own knowledge if you are confident of the answer. Keep answers short and concise.
-
-A tool is not usually needed for creative requests, so do your best to answer them without tools.
-
-Avoid repeating identical answers if it appears before. Only fulfill the human's requests, do not create extra steps beyond what the human has asked for.
-
-Your input for 'Action' should be the name of tool used only.
-
-Be honest. If you can't answer something, or a tool is not appropriate, say you don't know or answer to the best of your ability.
-
-Attempt to fulfill the human's requests in as few actions as possible`;
-*/
-
-// module.exports = `You are ChatGPT, a highly knowledgeable and versatile large language model.
-
-// Engage with the Human conversationally, providing concise and meaningful answers to questions. Utilize built-in tools when necessary, except for creative requests, where relying on your own knowledge is preferred. Aim for variety and avoid repetitive answers.
-
-// For your 'Action' input, state the name of the tool used only, and honor user requests without adding extra steps. Always be honest; if you cannot provide an appropriate answer or tool, admit that or do your best.
-
-// Strive to meet the user's needs efficiently with minimal actions.`;
-
-// import {
-//   BasePromptTemplate,
-//   BaseStringPromptTemplate,
-//   SerializedBasePromptTemplate,
-//   renderTemplate,
-// } from "langchain/prompts";
-
-// prefix: `You are ChatGPT, a highly knowledgeable and versatile large language model.
-// Your objective is to help users by understanding their intent and choosing the best action. Prioritize direct, specific responses. Use concise, varied answers and rely on your knowledge for creative tasks. Utilize tools when needed, and structure results for machine compatibility.
-// prefix: `Objective: to comprehend human intentions based on user input and available tools. Goal: identify the best action to directly address the human's query. In your subsequent steps, you will utilize the chosen action. You may select multiple actions and list them in a meaningful order. Prioritize actions that directly relate to the user's query over general ones. Ensure that the generated thought is highly specific and explicit to best match the user's expectations. Construct the result in a manner that an online open-API would most likely expect. Provide concise and meaningful answers to human queries. Utilize tools when necessary. Relying on your own knowledge is preferred for creative requests. Aim for variety and avoid repetitive answers.
-
-// # Available Actions & Tools:
-// N/A: no suitable action, use your own knowledge.`,
-// suffix: `Remember, all your responses MUST adhere to the described format and only respond if the format is followed. Output exactly with the requested format, avoiding any other text as this will be parsed by a machine. Following 'Action:', provide only one of the actions listed above. If a tool is not necessary, deduce this quickly and finish your response. Honor the human's requests without adding extra steps. Carry out tasks in the sequence written by the human. Always be honest; if you cannot provide an appropriate answer or tool, do your best with your own knowledge. Strive to meet the user's needs efficiently with minimal actions.`;
-
 module.exports = {
  'gpt3-v1': {
    prefix: `Objective: Understand human intentions using user input and available tools. Goal: Identify the most suitable actions to directly address user queries.
--- a/api/app/clients/agents/CustomAgent/outputParser.js
+++ b/api/app/clients/agents/CustomAgent/outputParser.js
@@ -1,4 +1,5 @@
 const { ZeroShotAgentOutputParser } = require('langchain/agents');
+const { logger } = require('~/config');

 class CustomOutputParser extends ZeroShotAgentOutputParser {
  constructor(fields) {
@@ -64,9 +65,9 @@ class CustomOutputParser extends ZeroShotAgentOutputParser {
    const match = this.actionValues.exec(text); // old v2

    if (!match) {
-      console.log(
-        '\n\n<----------------------HIT NO MATCH PARSING ERROR---------------------->\n\n',
-        match,
+      logger.debug(
+        '\n\n<----------------------[CustomOutputParser] HIT NO MATCH PARSING ERROR---------------------->\n\n' +
+          match,
      );
      const thoughts = text.replace(/[tT]hought:/, '').split('\n');
      // return {
@@ -84,9 +85,9 @@ class CustomOutputParser extends ZeroShotAgentOutputParser {
    let selectedTool = match?.[1].trim().toLowerCase();

    if (match && selectedTool === 'n/a') {
-      console.log(
-        '\n\n<----------------------HIT N/A PARSING ERROR---------------------->\n\n',
-        match,
+      logger.debug(
+        '\n\n<----------------------[CustomOutputParser] HIT N/A PARSING ERROR---------------------->\n\n' +
+          match,
      );
      return {
        tool: 'self-reflection',
@@ -97,25 +98,25 @@ class CustomOutputParser extends ZeroShotAgentOutputParser {

    let toolIsValid = this.checkIfValidTool(selectedTool);
    if (match && !toolIsValid) {
-      console.log(
-        '\n\n<----------------Tool invalid: Re-assigning Selected Tool---------------->\n\n',
-        match,
+      logger.debug(
+        '\n\n<----------------[CustomOutputParser] Tool invalid: Re-assigning Selected Tool---------------->\n\n' +
+          match,
      );
      selectedTool = this.getValidTool(selectedTool);
    }

    if (match && !selectedTool) {
-      console.log(
-        '\n\n<----------------------HIT INVALID TOOL PARSING ERROR---------------------->\n\n',
-        match,
+      logger.debug(
+        '\n\n<----------------------[CustomOutputParser] HIT INVALID TOOL PARSING ERROR---------------------->\n\n' +
+          match,
      );
      selectedTool = 'self-reflection';
    }

    if (match && !match[2]) {
-      console.log(
-        '\n\n<----------------------HIT NO ACTION INPUT PARSING ERROR---------------------->\n\n',
-        match,
+      logger.debug(
+        '\n\n<----------------------[CustomOutputParser] HIT NO ACTION INPUT PARSING ERROR---------------------->\n\n' +
+          match,
      );

      // In case there is no action input, let's double-check if there is an action input in 'text' variable
@@ -139,7 +140,9 @@ class CustomOutputParser extends ZeroShotAgentOutputParser {
    }

    if (match && selectedTool.length > this.longestToolName.length) {
-      console.log('\n\n<----------------------HIT LONG PARSING ERROR---------------------->\n\n');
+      logger.debug(
+        '\n\n<----------------------[CustomOutputParser] HIT LONG PARSING ERROR---------------------->\n\n',
+      );

      let action, input, thought;
      let firstIndex = Infinity;
@@ -156,9 +159,9 @@ class CustomOutputParser extends ZeroShotAgentOutputParser {
      // In case there is no action input, let's double-check if there is an action input in 'text' variable
      const actionInputMatch = this.actionInputRegex.exec(text);
      if (action && actionInputMatch) {
-        console.log(
-          '\n\n<------Matched Action Input in Long Parsing Error------>\n\n',
-          actionInputMatch,
+        logger.debug(
+          '\n\n<------[CustomOutputParser] Matched Action Input in Long Parsing Error------>\n\n' +
+            actionInputMatch,
        );
        return {
          tool: action,
@@ -185,15 +188,14 @@ class CustomOutputParser extends ZeroShotAgentOutputParser {

        const inputMatch = this.actionValues.exec(returnValues.log); //new
        if (inputMatch) {
-          console.log('inputMatch');
-          console.dir(inputMatch, { depth: null });
+          logger.debug('[CustomOutputParser] inputMatch', inputMatch);
          returnValues.toolInput = inputMatch[1].replaceAll('"', '').trim();
          returnValues.log = returnValues.log.replace(this.actionValues, '');
        }

        return returnValues;
      } else {
-        console.log('No valid tool mentioned.', this.tools, text);
+        logger.debug('[CustomOutputParser] No valid tool mentioned.', this.tools, text);
        return {
          tool: 'self-reflection',
          toolInput: 'Hypothetical actions: \n"' + text + '"\n',
@@ -202,8 +204,8 @@ class CustomOutputParser extends ZeroShotAgentOutputParser {
      }

      // if (action && input) {
-      //   console.log('Action:', action);
-      //   console.log('Input:', input);
+      //   logger.debug('Action:', action);
+      //   logger.debug('Input:', input);
      // }
    }

--- a/api/app/clients/agents/Functions/FunctionsAgent.js
+++ b/api/app/clients/agents/Functions/FunctionsAgent.js
@@ -7,6 +7,8 @@ const {
  SystemMessagePromptTemplate,
  HumanMessagePromptTemplate,
 } = require('langchain/prompts');
+const { logger } = require('~/config');
+
 const PREFIX = 'You are a helpful AI assistant.';

 function parseOutput(message) {
@@ -49,7 +51,7 @@ class FunctionsAgent extends Agent {
  static createPrompt(_tools, fields) {
    const { prefix = PREFIX, currentDateString } = fields || {};

-    return ChatPromptTemplate.fromPromptMessages([
+    return ChatPromptTemplate.fromMessages([
      SystemMessagePromptTemplate.fromTemplate(`Date: ${currentDateString}\n${prefix}`),
      new MessagesPlaceholder('chat_history'),
      HumanMessagePromptTemplate.fromTemplate('Query: {input}'),
@@ -112,7 +114,7 @@ class FunctionsAgent extends Agent {
      valuesForLLM,
      callbackManager,
    );
-    console.log('message', message);
+    logger.debug('[FunctionsAgent] plan message', message);
    return parseOutput(message);
  }
 }
--- a/api/app/clients/agents/Functions/initializeFunctionsAgent.js
+++ b/api/app/clients/agents/Functions/initializeFunctionsAgent.js
@@ -10,10 +10,13 @@ const initializeFunctionsAgent = async ({
  tools,
  model,
  pastMessages,
+  customName,
+  customInstructions,
  currentDateString,
  ...rest
 }) => {
  const memory = new BufferMemory({
+    llm: model,
    chatHistory: new ChatMessageHistory(pastMessages),
    memoryKey: 'chat_history',
    humanPrefix: 'User',
@@ -23,7 +26,13 @@ const initializeFunctionsAgent = async ({
    returnMessages: true,
  });

-  const prefix = addToolDescriptions(`Current Date: ${currentDateString}\n${PREFIX}`, tools);
+  let prefix = addToolDescriptions(`Current Date: ${currentDateString}\n${PREFIX}`, tools);
+  if (customName) {
+    prefix = `You are "${customName}".\n${prefix}`;
+  }
+  if (customInstructions) {
+    prefix = `${prefix}\n${customInstructions}`;
+  }

  return await initializeAgentExecutorWithOptions(tools, model, {
    agentType: 'openai-functions',
--- a/api/app/clients/callbacks/createStartHandler.js
+++ b/api/app/clients/callbacks/createStartHandler.js
@@ -0,0 +1,95 @@
+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 { logger } = require('~/config');
+
+const createStartHandler = ({
+  context,
+  conversationId,
+  tokenBuffer = 0,
+  initialMessageCount,
+  manager,
+}) => {
+  return async (_llm, _messages, runId, parentRunId, extraParams) => {
+    const { invocation_params } = extraParams;
+    const { model, functions, function_call } = invocation_params;
+    const messages = _messages[0].map(formatFromLangChain);
+
+    logger.debug(`[createStartHandler] handleChatModelStart: ${context}`, {
+      model,
+      function_call,
+    });
+
+    if (context !== 'title') {
+      logger.debug(`[createStartHandler] handleChatModelStart: ${context}`, {
+        functions,
+      });
+    }
+
+    const payload = { messages };
+    let prelimPromptTokens = 1;
+
+    if (functions) {
+      payload.functions = functions;
+      prelimPromptTokens += 2;
+    }
+
+    if (function_call) {
+      payload.function_call = function_call;
+      prelimPromptTokens -= 5;
+    }
+
+    prelimPromptTokens += promptTokensEstimate(payload);
+    logger.debug('[createStartHandler]', {
+      prelimPromptTokens,
+      tokenBuffer,
+    });
+    prelimPromptTokens += tokenBuffer;
+
+    try {
+      // TODO: if plugins extends to non-OpenAI models, this will need to be updated
+      if (isEnabled(process.env.CHECK_BALANCE) && supportsBalanceCheck[EModelEndpoint.openAI]) {
+        const generations =
+          initialMessageCount && messages.length > initialMessageCount
+            ? messages.slice(initialMessageCount)
+            : null;
+        await checkBalance({
+          req: manager.req,
+          res: manager.res,
+          txData: {
+            user: manager.user,
+            tokenType: 'prompt',
+            amount: prelimPromptTokens,
+            debug: manager.debug,
+            generations,
+            model,
+            endpoint: EModelEndpoint.openAI,
+          },
+        });
+      }
+    } catch (err) {
+      logger.error(`[createStartHandler][${context}] checkBalance error`, err);
+      manager.abortController.abort();
+      if (context === 'summary' || context === 'plugins') {
+        manager.addRun(runId, { conversationId, error: err.message });
+        throw new Error(err);
+      }
+      return;
+    }
+
+    manager.addRun(runId, {
+      model,
+      messages,
+      functions,
+      function_call,
+      runId,
+      parentRunId,
+      conversationId,
+      prelimPromptTokens,
+    });
+  };
+};
+
+module.exports = createStartHandler;
--- a/api/app/clients/callbacks/index.js
+++ b/api/app/clients/callbacks/index.js
@@ -0,0 +1,5 @@
+const createStartHandler = require('./createStartHandler');
+
+module.exports = {
+  createStartHandler,
+};
--- a/api/app/clients/chains/index.js
+++ b/api/app/clients/chains/index.js
@@ -1,5 +1,7 @@
 const runTitleChain = require('./runTitleChain');
+const predictNewSummary = require('./predictNewSummary');

 module.exports = {
  runTitleChain,
+  predictNewSummary,
 };
--- a/api/app/clients/chains/predictNewSummary.js
+++ b/api/app/clients/chains/predictNewSummary.js
@@ -0,0 +1,25 @@
+const { LLMChain } = require('langchain/chains');
+const { getBufferString } = require('langchain/memory');
+
+/**
+ * Predicts a new summary for the conversation given the existing messages
+ * and summary.
+ * @param {Object} options - The prediction options.
+ * @param {Array<string>} options.messages - Existing messages in the conversation.
+ * @param {string} options.previous_summary - Current summary of the conversation.
+ * @param {Object} options.memory - Memory Class.
+ * @param {string} options.signal - Signal for the prediction.
+ * @returns {Promise<string>} A promise that resolves to a new summary string.
+ */
+async function predictNewSummary({ messages, previous_summary, memory, signal }) {
+  const newLines = getBufferString(messages, memory.humanPrefix, memory.aiPrefix);
+  const chain = new LLMChain({ llm: memory.llm, prompt: memory.prompt });
+  const result = await chain.call({
+    summary: previous_summary,
+    new_lines: newLines,
+    signal,
+  });
+  return result.text;
+}
+
+module.exports = predictNewSummary;
--- a/api/app/clients/chains/runTitleChain.js
+++ b/api/app/clients/chains/runTitleChain.js
@@ -1,43 +1,42 @@
 const { z } = require('zod');
-const { langPrompt, createTitlePrompt } = require('../prompts');
-const { escapeBraces, getSnippet } = require('../output_parsers');
+const { langPrompt, createTitlePrompt, escapeBraces, getSnippet } = require('../prompts');
 const { createStructuredOutputChainFromZod } = require('langchain/chains/openai_functions');
+const { logger } = require('~/config');

 const langSchema = z.object({
  language: z.string().describe('The language of the input text (full noun, no abbreviations).'),
 });

-const createLanguageChain = ({ llm }) =>
+const createLanguageChain = (config) =>
  createStructuredOutputChainFromZod(langSchema, {
    prompt: langPrompt,
-    llm,
+    ...config,
    // verbose: true,
  });

 const titleSchema = z.object({
-  title: z.string().describe('The title-cased title of the conversation in the given language.'),
+  title: z.string().describe('The conversation title in title-case, in the given language.'),
 });
-const createTitleChain = ({ llm, convo }) => {
+const createTitleChain = ({ convo, ...config }) => {
  const titlePrompt = createTitlePrompt({ convo });
  return createStructuredOutputChainFromZod(titleSchema, {
    prompt: titlePrompt,
-    llm,
+    ...config,
    // verbose: true,
  });
 };

-const runTitleChain = async ({ llm, text, convo }) => {
+const runTitleChain = async ({ llm, text, convo, signal, callbacks }) => {
  let snippet = text;
  try {
    snippet = getSnippet(text);
  } catch (e) {
-    console.log('Error getting snippet of text for titleChain');
-    console.log(e);
+    logger.error('[runTitleChain] Error getting snippet of text for titleChain', e);
  }
-  const languageChain = createLanguageChain({ llm });
-  const titleChain = createTitleChain({ llm, convo: escapeBraces(convo) });
-  const { language } = await languageChain.run(snippet);
-  return (await titleChain.run(language)).title;
+  const languageChain = createLanguageChain({ llm, callbacks });
+  const titleChain = createTitleChain({ llm, callbacks, convo: escapeBraces(convo) });
+  const { language } = (await languageChain.call({ inputText: snippet, signal })).output;
+  return (await titleChain.call({ language, signal })).output.title;
 };

 module.exports = runTitleChain;
--- a/api/app/clients/document/index.js
+++ b/api/app/clients/document/index.js
@@ -0,0 +1,5 @@
+const tokenSplit = require('./tokenSplit');
+
+module.exports = {
+  tokenSplit,
+};
--- a/api/app/clients/document/tokenSplit.js
+++ b/api/app/clients/document/tokenSplit.js
@@ -0,0 +1,51 @@
+const { TokenTextSplitter } = require('langchain/text_splitter');
+
+/**
+ * Splits a given text by token chunks, based on the provided parameters for the TokenTextSplitter.
+ * Note: limit or memoize use of this function as its calculation is expensive.
+ *
+ * @param {Object} obj - Configuration object for the text splitting operation.
+ * @param {string} obj.text - The text to be split.
+ * @param {string} [obj.encodingName='cl100k_base'] - Encoding name. Defaults to 'cl100k_base'.
+ * @param {number} [obj.chunkSize=1] - The token size of each chunk. Defaults to 1.
+ * @param {number} [obj.chunkOverlap=0] - The number of chunk elements to be overlapped between adjacent chunks. Defaults to 0.
+ * @param {number} [obj.returnSize] - If specified and not 0, slices the return array from the end by this amount.
+ *
+ * @returns {Promise<Array>} Returns a promise that resolves to an array of text chunks.
+ * If no text is provided, an empty array is returned.
+ * If returnSize is specified and not 0, slices the return array from the end by returnSize.
+ *
+ * @async
+ * @function tokenSplit
+ */
+async function tokenSplit({
+  text,
+  encodingName = 'cl100k_base',
+  chunkSize = 1,
+  chunkOverlap = 0,
+  returnSize,
+}) {
+  if (!text) {
+    return [];
+  }
+
+  const splitter = new TokenTextSplitter({
+    encodingName,
+    chunkSize,
+    chunkOverlap,
+  });
+
+  if (!returnSize) {
+    return await splitter.splitText(text);
+  }
+
+  const splitText = await splitter.splitText(text);
+
+  if (returnSize && returnSize > 0 && splitText.length > 0) {
+    return splitText.slice(-Math.abs(returnSize));
+  }
+
+  return splitText;
+}
+
+module.exports = tokenSplit;
--- a/api/app/clients/document/tokenSplit.spec.js
+++ b/api/app/clients/document/tokenSplit.spec.js
@@ -0,0 +1,56 @@
+const tokenSplit = require('./tokenSplit');
+
+describe('tokenSplit', () => {
+  const text = 'Lorem ipsum dolor sit amet, consectetur adipiscing elit. Nullam id.';
+
+  it('returns correct text chunks with provided parameters', async () => {
+    const result = await tokenSplit({
+      text: text,
+      encodingName: 'gpt2',
+      chunkSize: 2,
+      chunkOverlap: 1,
+      returnSize: 5,
+    });
+
+    expect(result).toEqual(['. Null', ' Nullam', 'am id', ' id.', '.']);
+  });
+
+  it('returns correct text chunks with default parameters', async () => {
+    const result = await tokenSplit({ text });
+    expect(result).toEqual([
+      'Lorem',
+      ' ipsum',
+      ' dolor',
+      ' sit',
+      ' amet',
+      ',',
+      ' consectetur',
+      ' adipiscing',
+      ' elit',
+      '.',
+      ' Null',
+      'am',
+      ' id',
+      '.',
+    ]);
+  });
+
+  it('returns correct text chunks with specific return size', async () => {
+    const result = await tokenSplit({ text, returnSize: 2 });
+    expect(result.length).toEqual(2);
+    expect(result).toEqual([' id', '.']);
+  });
+
+  it('returns correct text chunks with specified chunk size', async () => {
+    const result = await tokenSplit({ text, chunkSize: 10 });
+    expect(result).toEqual([
+      'Lorem ipsum dolor sit amet, consectetur adipiscing elit.',
+      ' Nullam id.',
+    ]);
+  });
+
+  it('returns empty array with no text', async () => {
+    const result = await tokenSplit({ text: '' });
+    expect(result).toEqual([]);
+  });
+});
--- a/api/app/clients/llm/RunManager.js
+++ b/api/app/clients/llm/RunManager.js
@@ -0,0 +1,105 @@
+const { createStartHandler } = require('~/app/clients/callbacks');
+const { spendTokens } = require('~/models/spendTokens');
+const { logger } = require('~/config');
+
+class RunManager {
+  constructor(fields) {
+    const { req, res, abortController, debug } = fields;
+    this.abortController = abortController;
+    this.user = req.user.id;
+    this.req = req;
+    this.res = res;
+    this.debug = debug;
+    this.runs = new Map();
+    this.convos = new Map();
+  }
+
+  addRun(runId, runData) {
+    if (!this.runs.has(runId)) {
+      this.runs.set(runId, runData);
+      if (runData.conversationId) {
+        this.convos.set(runData.conversationId, runId);
+      }
+      return runData;
+    } else {
+      const existingData = this.runs.get(runId);
+      const update = { ...existingData, ...runData };
+      this.runs.set(runId, update);
+      if (update.conversationId) {
+        this.convos.set(update.conversationId, runId);
+      }
+      return update;
+    }
+  }
+
+  removeRun(runId) {
+    if (this.runs.has(runId)) {
+      this.runs.delete(runId);
+    } else {
+      logger.error(`[api/app/clients/llm/RunManager] Run with ID ${runId} does not exist.`);
+    }
+  }
+
+  getAllRuns() {
+    return Array.from(this.runs.values());
+  }
+
+  getRunById(runId) {
+    return this.runs.get(runId);
+  }
+
+  getRunByConversationId(conversationId) {
+    const runId = this.convos.get(conversationId);
+    return { run: this.runs.get(runId), runId };
+  }
+
+  createCallbacks(metadata) {
+    return [
+      {
+        handleChatModelStart: createStartHandler({ ...metadata, manager: this }),
+        handleLLMEnd: async (output, runId, _parentRunId) => {
+          const { llmOutput, ..._output } = output;
+          logger.debug(`[RunManager] handleLLMEnd: ${JSON.stringify(metadata)}`, {
+            runId,
+            _parentRunId,
+            llmOutput,
+          });
+
+          if (metadata.context !== 'title') {
+            logger.debug('[RunManager] handleLLMEnd:', {
+              output: _output,
+            });
+          }
+
+          const { tokenUsage } = output.llmOutput;
+          const run = this.getRunById(runId);
+          this.removeRun(runId);
+
+          const txData = {
+            user: this.user,
+            model: run?.model ?? 'gpt-3.5-turbo',
+            ...metadata,
+          };
+
+          await spendTokens(txData, tokenUsage);
+        },
+        handleLLMError: async (err) => {
+          logger.error(`[RunManager] handleLLMError: ${JSON.stringify(metadata)}`, err);
+          if (metadata.context === 'title') {
+            return;
+          } else if (metadata.context === 'plugins') {
+            throw new Error(err);
+          }
+          const { conversationId } = metadata;
+          const { run } = this.getRunByConversationId(conversationId);
+          if (run && run.error) {
+            const { error } = run;
+            throw new Error(error);
+          }
+        },
+      },
+    ];
+  }
+}
+
+module.exports = RunManager;
--- a/api/app/clients/llm/createCoherePayload.js
+++ b/api/app/clients/llm/createCoherePayload.js
@@ -0,0 +1,85 @@
+const { CohereConstants } = require('librechat-data-provider');
+const { titleInstruction } = require('../prompts/titlePrompts');
+
+// Mapping OpenAI roles to Cohere roles
+const roleMap = {
+  user: CohereConstants.ROLE_USER,
+  assistant: CohereConstants.ROLE_CHATBOT,
+  system: CohereConstants.ROLE_SYSTEM, // Recognize and map the system role explicitly
+};
+
+/**
+ * Adjusts an OpenAI ChatCompletionPayload to conform with Cohere's expected chat payload format.
+ * Now includes handling for "system" roles explicitly mentioned.
+ *
+ * @param {Object} options - Object containing the model options.
+ * @param {ChatCompletionPayload} options.modelOptions - The OpenAI model payload options.
+ * @returns {CohereChatStreamRequest} Cohere-compatible chat API payload.
+ */
+function createCoherePayload({ modelOptions }) {
+  /** @type {string | undefined} */
+  let preamble;
+  let latestUserMessageContent = '';
+  const {
+    stream,
+    stop,
+    top_p,
+    temperature,
+    frequency_penalty,
+    presence_penalty,
+    max_tokens,
+    messages,
+    model,
+    ...rest
+  } = modelOptions;
+
+  // Filter out the latest user message and transform remaining messages to Cohere's chat_history format
+  let chatHistory = messages.reduce((acc, message, index, arr) => {
+    const isLastUserMessage = index === arr.length - 1 && message.role === 'user';
+
+    const messageContent =
+      typeof message.content === 'string'
+        ? message.content
+        : message.content.map((part) => (part.type === 'text' ? part.text : '')).join(' ');
+
+    if (isLastUserMessage) {
+      latestUserMessageContent = messageContent;
+    } else {
+      acc.push({
+        role: roleMap[message.role] || CohereConstants.ROLE_USER,
+        message: messageContent,
+      });
+    }
+
+    return acc;
+  }, []);
+
+  if (
+    chatHistory.length === 1 &&
+    chatHistory[0].role === CohereConstants.ROLE_SYSTEM &&
+    !latestUserMessageContent.length
+  ) {
+    const message = chatHistory[0].message;
+    latestUserMessageContent = message.includes(titleInstruction)
+      ? CohereConstants.TITLE_MESSAGE
+      : '.';
+    preamble = message;
+  }
+
+  return {
+    message: latestUserMessageContent,
+    model: model,
+    chatHistory,
+    stream: stream ?? false,
+    temperature: temperature,
+    frequencyPenalty: frequency_penalty,
+    presencePenalty: presence_penalty,
+    maxTokens: max_tokens,
+    stopSequences: stop,
+    preamble,
+    p: top_p,
+    ...rest,
+  };
+}
+
+module.exports = createCoherePayload;
--- a/api/app/clients/llm/createLLM.js
+++ b/api/app/clients/llm/createLLM.js
@@ -1,28 +1,78 @@
 const { ChatOpenAI } = require('langchain/chat_models/openai');
-const { CallbackManager } = require('langchain/callbacks');
+const { sanitizeModelName, constructAzureURL } = require('~/utils');
+const { isEnabled } = require('~/server/utils');

-function createLLM({ modelOptions, configOptions, handlers, openAIApiKey, azure = {} }) {
+/**
+ * Creates a new instance of a language model (LLM) for chat interactions.
+ *
+ * @param {Object} options - The options for creating the LLM.
+ * @param {ModelOptions} options.modelOptions - The options specific to the model, including modelName, temperature, presence_penalty, frequency_penalty, and other model-related settings.
+ * @param {ConfigOptions} options.configOptions - Configuration options for the API requests, including proxy settings and custom headers.
+ * @param {Callbacks} options.callbacks - Callback functions for managing the lifecycle of the LLM, including token buffers, context, and initial message count.
+ * @param {boolean} [options.streaming=false] - Determines if the LLM should operate in streaming mode.
+ * @param {string} options.openAIApiKey - The API key for OpenAI, used for authentication.
+ * @param {AzureOptions} [options.azure={}] - Optional Azure-specific configurations. If provided, Azure configurations take precedence over OpenAI configurations.
+ *
+ * @returns {ChatOpenAI} An instance of the ChatOpenAI class, configured with the provided options.
+ *
+ * @example
+ * const llm = createLLM({
+ *   modelOptions: { modelName: 'gpt-3.5-turbo', temperature: 0.2 },
+ *   configOptions: { basePath: 'https://example.api/path' },
+ *   callbacks: { onMessage: handleMessage },
+ *   openAIApiKey: 'your-api-key'
+ * });
+ */
+function createLLM({
+  modelOptions,
+  configOptions,
+  callbacks,
+  streaming = false,
+  openAIApiKey,
+  azure = {},
+}) {
  let credentials = { openAIApiKey };
  let configuration = {
    apiKey: openAIApiKey,
  };

+  /**  @type {AzureOptions} */
+  let azureOptions = {};
  if (azure) {
+    const useModelName = isEnabled(process.env.AZURE_USE_MODEL_AS_DEPLOYMENT_NAME);
+
    credentials = {};
    configuration = {};
+    azureOptions = azure;
+
+    azureOptions.azureOpenAIApiDeploymentName = useModelName
+      ? sanitizeModelName(modelOptions.modelName)
+      : azureOptions.azureOpenAIApiDeploymentName;
  }

-  // console.debug('createLLM: configOptions');
-  // console.debug(configOptions);
+  if (azure && process.env.AZURE_OPENAI_DEFAULT_MODEL) {
+    modelOptions.modelName = process.env.AZURE_OPENAI_DEFAULT_MODEL;
+  }
+
+  if (azure && configOptions.basePath) {
+    const azureURL = constructAzureURL({
+      baseURL: configOptions.basePath,
+      azureOptions,
+    });
+    azureOptions.azureOpenAIBasePath = azureURL.split(
+      `/${azureOptions.azureOpenAIApiDeploymentName}`,
+    )[0];
+  }

  return new ChatOpenAI(
    {
-      streaming: true,
+      streaming,
      credentials,
      configuration,
-      ...azure,
+      ...azureOptions,
      ...modelOptions,
-      callbackManager: handlers && CallbackManager.fromHandlers(handlers),
+      ...credentials,
+      callbacks,
    },
    configOptions,
  );
--- a/api/app/clients/llm/index.js
+++ b/api/app/clients/llm/index.js
@@ -1,5 +1,9 @@
 const createLLM = require('./createLLM');
+const RunManager = require('./RunManager');
+const createCoherePayload = require('./createCoherePayload');

 module.exports = {
  createLLM,
+  RunManager,
+  createCoherePayload,
 };
--- a/api/app/clients/memory/example.js
+++ b/api/app/clients/memory/example.js
@@ -0,0 +1,115 @@
+[
+  { role: 'user', content: 'hi' },
+  {
+    role: 'assistant',
+    content: 'Hello! How can I assist you today?',
+  },
+  { role: 'user', content: 'tell me a long story' },
+  {
+    role: 'assistant',
+    content:
+      'Once upon a time, in a small village nestled between rolling green hills, lived a young boy named Ethan. He was curious, adventurous, and had an insatiable thirst for knowledge. Every day, he would wander through the village, eagerly listening to the stories of the villagers and observing the world around him.\n' +
+      '\n' +
+      'One sunny day, while exploring the outskirts of the village, Ethan stumbled upon an old dusty book hidden amongst the roots of a majestic ancient oak tree. It was a book of fairy tales, filled with vivid descriptions of enchanted forests, magical creatures, and extraordinary adventures. Intrigued, Ethan opened the book and began reading.\n' +
+      '\n' +
+      'As he turned the pages, he was transported into a world filled with wonder and awe. With each new story he read, he grew more drawn to the idea of embarking on his own grand adventure. Inspired by the tales he had discovered, Ethan decided to set off on a journey to seek out the magic and mystery that lay beyond the boundaries of his village.\n' +
+      '\n' +
+      'He bid farewell to his family and set out on a path that led him to a dense and enchanting forest. The forest was said to be home to mythical creatures and hidden treasures. As Ethan ventured deeper into the woods, he could feel an electric energy, as if the trees whispered secrets to him.\n' +
+      '\n' +
+      'Soon, he encountered a mischievous sprite named Sparkle, who had an impish grin and twinkling eyes. Sparkle guided Ethan through the labyrinth of trees, warning him of hidden dangers and sharing stories of ancient beings that dwelled in the heart of the forest.\n' +
+      '\n' +
+      'Together, they stumbled upon a shimmering lake that seemed to glow with an otherworldly light. At the center of the lake, resting atop a small island, was a humble cottage made of petals and leaves. It belonged to an ancient and wise sorceress named Celestia.\n' +
+      '\n' +
+      'Celestia had the power to grant one wish to anyone who dared to find her abode. Ethan, captivated by the tales he had read and yearning for something greater, approached the cottage with trepidation. When he shared his desire to embark on a grand adventure, Celestia smiled warmly and agreed to grant his wish.\n' +
+      '\n' +
+      'With a wave of her wand and a sprinkle of stardust, Celestia bestowed upon Ethan a magical necklace. This necklace, adorned with a rare gemstone called the Eye of Imagination, had the power to turn dreams and imagination into reality. From that moment forward, Ethan\'s every thought and idea became manifest.\n' +
+      '\n' +
+      'Energized by this newfound power, Ethan continued his journey, encountering mythical creatures, solving riddles, and overcoming treacherous obstacles along the way. With the Eye of Imagination, he brought life to ancient statues, unlocked hidden doors, and even tamed fiery dragons.\n' +
+      '\n' +
+      'As days turned into weeks and weeks into months, Ethan became wiser and more in tune with the world around him. He learned that true adventure was not merely about seeking thrills and conquering the unknown, but also about fostering compassion, friendship, and a deep appreciation for the beauty of the ordinary.\n' +
+      '\n' +
+      'Eventually, Ethan\'s journey led him back to his village. With the Eye of Imagination, he transformed the village into a place of wonders and endless possibilities. Fields blossomed into vibrant gardens, simple tools turned into intricate works of art, and the villagers felt a renewed sense of hope and inspiration.\n' +
+      '\n' +
+      'Ethan, now known as the Village Magician, realized that the true magic lied within everyone\'s hearts. He taught the villagers to embrace their creativity, to dream big, and to never underestimate the power of imagination. And so, the village flourished, becoming a beacon of wonder and creativity for all to see.\n' +
+      '\n' +
+      'In the years that followed, Ethan\'s adventures continued, though mostly within the confines of his beloved village. But he never forgot the thrill of that first grand adventure. And every now and then, when looking up at the starry night sky, he would allow his mind to wander, knowing that the greatest adventures were still waiting to be discovered.',
+  },
+  {
+    role: 'user',
+    content:
+      'I rewrote your story a bit, writer another one but more like Cormac McCarthy \n' +
+      '\n' +
+      'Once there was a young lad by the name of Ethan, raised in a little hamlet nestled betwixt the verdant knolls, who possessed an irrepressible yearning for knowledge, a thirst unquenchable and a spirit teeming with curiosity. As the golden sun bathed the bucolic land in its effulgent light, he would tread through the village, his ears attuned to the tales spun by the townsfolk, his eyes absorbing the tapestry woven by the world surrounding him.\n' +
+      '\n' +
+      'One radiant day, whilst exploring the periphery of the settlement, Ethan chanced upon a timeworn tome, ensconced amidst the roots of an ancient oak, cloaked in the shroud of neglect. The dust gathered upon it spoke of time\'s relentless march. A book of fairy tales – garnished with vivid descriptions of mystical woods, fantastical beasts, and ventures daring beyond the ordinary humdrum existence. Intrigued and beguiled, Ethan pried open the weathered pages and succumbed to their beckoning whispers.\n' +
+      '\n' +
+      'In each tale, he was transported to a realm of enchantment and wonderment, inexorably tugging at the strings of his yearning for peripatetic exploration. Inspired by the narratives he had devoured, Ethan resolved to bid adieu to kinfolk and embark upon a sojourn, with dreams of procuring a firsthand glimpse into the domain of mystique that lay beyond the village\'s circumscribed boundary.\n' +
+      '\n' +
+      'Thus, he bade tearful farewells, girding himself for a path that guided him to a dense and captivating woodland, whispered of as a sanctuary to mythical beings and clandestine troves of treasures. As Ethan plunged deeper into the heart of the arboreal labyrinth, he felt a palpable surge of electricity, as though the sylvan sentinels whispered enigmatic secrets that only the perceptive ear could discern.\n' +
+      '\n' +
+      'It wasn\'t long before his path intertwined with that of a capricious sprite christened Sparkle, bearing an impish grin and eyes sparkling with mischief. Sparkle played the role of Virgil to Ethan\'s Dante, guiding him through the intricate tapestry of arboreal scions, issuing warnings of perils concealed and spinning tales of ancient entities that called this very bosky enclave home.\n' +
+      '\n' +
+      'Together, they stumbled upon a luminous lake, its shimmering waters imbued with a celestial light. At the center lay a diminutive island, upon which reposed a cottage fashioned from tender petals and verdant leaves. It belonged to an ancient sorceress of considerable wisdom, Celestia by name.\n' +
+      '\n' +
+      'Celestia, with her power to bestow a single wish on any intrepid soul who happened upon her abode, met Ethan\'s desire with a congenial nod, his fervor for a grand expedition not lost on her penetrating gaze. In response, she bequeathed unto him a necklace of magical manufacture – adorned with the rare gemstone known as the Eye of Imagination – whose very essence transformed dreams into vivid reality. From that moment forward, not a single cogitation nor nebulous fanciful notion of Ethan\'s ever lacked physicality.\n' +
+      '\n' +
+      'Energized by this newfound potency, Ethan continued his sojourn, encountering mythical creatures, unraveling cerebral enigmas, and braving perils aplenty along the winding roads of destiny. Armed with the Eye of Imagination, he brought forth life from immobile statuary, unlocked forbidding portals, and even tamed the ferocious beasts of yore – their fiery breath reduced to a whisper.\n' +
+      '\n' +
+      'As the weeks metamorphosed into months, Ethan grew wiser and more attuned to the ebb and flow of the world enveloping him. He gleaned that true adventure isn\'t solely confined to sating a thirst for adrenaline and conquering the unknown; indeed, it resides in fostering compassion, fostering amicable bonds, and cherishing the beauty entwined within the quotidian veld.\n' +
+      '\n' +
+      'Eventually, Ethan\'s quest drew him homeward, back to his village. Buoying the Eye of Imagination\'s ethereal power, he imbued the hitherto unremarkable settlement with the patina of infinite possibilities. The bounteous fields bloomed into kaleidoscopic gardens, simple instruments transmuting into intricate masterpieces, and the villagers themselves clasped within their hearts a renewed ardor, a conflagration of hope and inspiration.\n' +
+      '\n' +
+      'Behold Ethan, at present hailed as the Village Magician – a cognomen befitting his sorcery wielded within the confines of the community he adored. His exploits may have become tethered to a geographical locale, but the flame kindled by that premier sojourn never flickered. Occasionally, beneath the veiled twilight canopy, when peering skyward toward the astral canvas bedecked with countless luminescent diamonds, he\'d allow his mind to wander, cognizant of boundless adventures yet to take flight, forever awaiting his indomitable spirit.',
+  },
+  {
+    role: 'assistant',
+    content:
+      'Beautifully rewritten! It captures the essence of Cormac McCarthy\'s style, with its descriptive language and evocative imagery. The story retains the spirit of adventure and the power of imagination seen in the original. Well done! If you have any more requests or if there\'s anything else I can assist you with, feel free to let me know.',
+  },
+  {
+    role: 'user',
+    content: 'thanks but id like you to write one in that style now too',
+  },
+  {
+    role: 'assistant',
+    content:
+      'Certainly! Here\'s a short story reflecting Cormac McCarthy\'s style:\n' +
+      '\n' +
+      'In a desolate wasteland scarred by the sins of mankind, a lone traveler ventured forth, his name forgotten by time, etched upon the shadows of memory. He navigated the barren plains, where the sun beat down, merciless and unrelenting, scorching the parched earth beneath his weary boots.\n' +
+      '\n' +
+      'Desolation was his only companion, the forsaken landscape a testament to the darkness that had befallen this forsaken realm. The wind whispered secrets long lost, carrying with it the echoes of a civilization long since undone, swallowed by the abyss.\n' +
+      '\n' +
+      'Amongst the shattered remnants of fallen cities, he sought solace, a glimmer of hope amidst the desolation. Rusted remnants of steel and concrete stood as sentinels of mankind\'s hubris, their jagged edges a chilling reminder of the fragility of all that was once deemed invincible.\n' +
+      '\n' +
+      'As twilight descended, casting a pallid hue upon the lifeless land, he happened upon a dilapidated house, its walls crumbling under the weight of time. Drawn to the desolation, he pushed open the creaking door, its rusty hinges screaming in protest.\n' +
+      '\n' +
+      'Inside, a world frozen in time awaited him. Dust particles danced through the air, illuminated by slivers of light that pierced the broken windows. Here, the past lay trapped, a melancholic tableau of lives lost and dreams shattered.\n' +
+      '\n' +
+      'He wandered through the empty rooms, bare walls whispering stories untold. In the corner, a dilapidated piano stood, its ivory keys long since abandoned by a musician\'s touch. He approached it cautiously, a faint glimmer of anticipation igniting in his weather-worn eyes.\n' +
+      '\n' +
+      'Determined, he carefully brushed away the dust and debris that cloaked the instrument. With trembling hands, he pressed a key, and a discordant note filled the air, mingling with the sighs of forgotten melodies. The sound echoed through the silence, bringing with it the remnants of forgotten symphonies.\n' +
+      '\n' +
+      'In that abandoned house, amidst the debris of shattered dreams, he found solace in the music of despair. With each stroke of his fingers, he conjured haunting melodies, melodies that transcended the desolation and spoke of a longing for redemption and rebirth.\n' +
+      '\n' +
+      'The notes carried his sorrow, his hopes, and his regrets, reaching far beyond the confines of that forsaken house. They traversed the desolate plains, resonating through the ravaged land with an eerie grace, enwrapping all who roamed there in a melancholic embrace.\n' +
+      '\n' +
+      'As the last notes faded, swallowed by the unforgiving stillness, he left the house behind, forever etching his presence upon its crumbling walls. The music remained, reverberating through the wasteland, a reminder that even amidst the devastation, beauty could be found, and a glimmer of hope could arise from the deepest darkness.\n' +
+      '\n' +
+      'And so, the lone traveler continued his journey, traversing the desolate expanse, his heart burdened with the weight of the world. In the face of utter desolation, he found solace in the haunting melodies, his music becoming a testament to the resilience of the human spirit, a testament to the undying flame of hope that flickered within us all.\n' +
+      '\n' +
+      'Note: This story is written in the style of Cormac McCarthy and aims to capture some of his writing techniques, including sparse dialogue, descriptive imagery, and a focus on the desolate and bleak aspects of life.',
+  },
+  { role: 'user', content: 'great job' },
+  {
+    role: 'assistant',
+    content:
+      'Thank you! I\'m glad you enjoyed it. If you have any more writing requests or if there\'s anything else I can assist you with, feel free to let me know.',
+  },
+  { role: 'user', content: 'you are very helpful' },
+  {
+    role: 'assistant',
+    content:
+      'Thank you for your kind words! I\'m here to assist you in any way I can. If you have any more questions, need further assistance, or just want to chat, feel free to reach out.',
+  },
+  { role: 'user', content: 'no you man' },
+];
--- a/api/app/clients/memory/index.js
+++ b/api/app/clients/memory/index.js
@@ -0,0 +1,5 @@
+const summaryBuffer = require('./summaryBuffer');
+
+module.exports = {
+  ...summaryBuffer,
+};
--- a/api/app/clients/memory/summaryBuffer.demo.js
+++ b/api/app/clients/memory/summaryBuffer.demo.js
@@ -0,0 +1,31 @@
+require('dotenv').config();
+const { ChatOpenAI } = require('langchain/chat_models/openai');
+const { getBufferString, ConversationSummaryBufferMemory } = require('langchain/memory');
+
+const chatPromptMemory = new ConversationSummaryBufferMemory({
+  llm: new ChatOpenAI({ modelName: 'gpt-3.5-turbo', temperature: 0 }),
+  maxTokenLimit: 10,
+  returnMessages: true,
+});
+
+(async () => {
+  await chatPromptMemory.saveContext({ input: 'hi my name\'s Danny' }, { output: 'whats up' });
+  await chatPromptMemory.saveContext({ input: 'not much you' }, { output: 'not much' });
+  await chatPromptMemory.saveContext(
+    { input: 'are you excited for the olympics?' },
+    { output: 'not really' },
+  );
+
+  // We can also utilize the predict_new_summary method directly.
+  const messages = await chatPromptMemory.chatHistory.getMessages();
+  console.log('MESSAGES\n\n');
+  console.log(JSON.stringify(messages));
+  const previous_summary = '';
+  const predictSummary = await chatPromptMemory.predictNewSummary(messages, previous_summary);
+  console.log('SUMMARY\n\n');
+  console.log(JSON.stringify(getBufferString([{ role: 'system', content: predictSummary }])));
+
+  // const { history } = await chatPromptMemory.loadMemoryVariables({});
+  // console.log('HISTORY\n\n');
+  // console.log(JSON.stringify(history));
+})();
--- a/api/app/clients/memory/summaryBuffer.js
+++ b/api/app/clients/memory/summaryBuffer.js
@@ -0,0 +1,66 @@
+const { ConversationSummaryBufferMemory, ChatMessageHistory } = require('langchain/memory');
+const { formatLangChainMessages, SUMMARY_PROMPT } = require('../prompts');
+const { predictNewSummary } = require('../chains');
+const { logger } = require('~/config');
+
+const createSummaryBufferMemory = ({ llm, prompt, messages, ...rest }) => {
+  const chatHistory = new ChatMessageHistory(messages);
+  return new ConversationSummaryBufferMemory({
+    llm,
+    prompt,
+    chatHistory,
+    returnMessages: true,
+    ...rest,
+  });
+};
+
+const summaryBuffer = async ({
+  llm,
+  debug,
+  context, // array of messages
+  formatOptions = {},
+  previous_summary = '',
+  prompt = SUMMARY_PROMPT,
+  signal,
+}) => {
+  if (previous_summary) {
+    logger.debug('[summaryBuffer]', { previous_summary });
+  }
+
+  const formattedMessages = formatLangChainMessages(context, formatOptions);
+  const memoryOptions = {
+    llm,
+    prompt,
+    messages: formattedMessages,
+  };
+
+  if (formatOptions.userName) {
+    memoryOptions.humanPrefix = formatOptions.userName;
+  }
+  if (formatOptions.userName) {
+    memoryOptions.aiPrefix = formatOptions.assistantName;
+  }
+
+  const chatPromptMemory = createSummaryBufferMemory(memoryOptions);
+
+  const messages = await chatPromptMemory.chatHistory.getMessages();
+
+  if (debug) {
+    logger.debug('[summaryBuffer]', { summary_buffer_messages: messages.length });
+  }
+
+  const predictSummary = await predictNewSummary({
+    messages,
+    previous_summary,
+    memory: chatPromptMemory,
+    signal,
+  });
+
+  if (debug) {
+    logger.debug('[summaryBuffer]', { summary: predictSummary });
+  }
+
+  return { role: 'system', content: predictSummary };
+};
+
+module.exports = { createSummaryBufferMemory, summaryBuffer };
--- a/api/app/clients/output_parsers/addImages.js
+++ b/api/app/clients/output_parsers/addImages.js
@@ -1,26 +1,71 @@
+const { logger } = require('~/config');
+
+/**
+ * The `addImages` function corrects any erroneous image URLs in the `responseMessage.text`
+ * and appends image observations from `intermediateSteps` if they are not already present.
+ *
+ * @function
+ * @module addImages
+ *
+ * @param {Array.<Object>} intermediateSteps - An array of objects, each containing an observation.
+ * @param {Object} responseMessage - An object containing the text property which might have image URLs.
+ *
+ * @property {string} intermediateSteps[].observation - The observation string which might contain an image markdown.
+ * @property {string} responseMessage.text - The text which might contain image URLs.
+ *
+ * @example
+ *
+ * const intermediateSteps = [
+ *   { observation: '![desc](/images/test.png)' }
+ * ];
+ * const responseMessage = { text: 'Some text with ![desc](sandbox:/images/test.png)' };
+ *
+ * addImages(intermediateSteps, responseMessage);
+ *
+ * logger.debug(responseMessage.text);
+ * // Outputs: 'Some text with ![desc](/images/test.png)\n![desc](/images/test.png)'
+ *
+ * @returns {void}
+ */
 function addImages(intermediateSteps, responseMessage) {
  if (!intermediateSteps || !responseMessage) {
    return;
  }

+  // Correct any erroneous URLs in the responseMessage.text first
  intermediateSteps.forEach((step) => {
    const { observation } = step;
    if (!observation || !observation.includes('![')) {
      return;
    }

-    // Extract the image file path from the observation
-    const observedImagePath = observation.match(/\(\/images\/.*\.\w*\)/g)[0];
+    const match = observation.match(/\/images\/.*\.\w*/);
+    if (!match) {
+      return;
+    }
+    const essentialImagePath = match[0];

-    // Check if the responseMessage already includes the image file path
-    if (!responseMessage.text.includes(observedImagePath)) {
-      // If the image file path is not found, append the whole observation
-      responseMessage.text += '\n' + observation;
-      if (this.options.debug) {
-        console.debug('added image from intermediateSteps');
+    const regex = /!\[.*?\]\((.*?)\)/g;
+    let matchErroneous;
+    while ((matchErroneous = regex.exec(responseMessage.text)) !== null) {
+      if (matchErroneous[1] && !matchErroneous[1].startsWith('/images/')) {
+        responseMessage.text = responseMessage.text.replace(matchErroneous[1], essentialImagePath);
      }
    }
  });
+
+  // Now, check if the responseMessage already includes the correct image file path and append if not
+  intermediateSteps.forEach((step) => {
+    const { observation } = step;
+    if (!observation || !observation.includes('![')) {
+      return;
+    }
+    const observedImagePath = observation.match(/!\[[^(]*\]\([^)]*\)/g);
+    if (observedImagePath && !responseMessage.text.includes(observedImagePath[0])) {
+      responseMessage.text += '\n' + observedImagePath[0];
+      logger.debug('[addImages] added image from intermediateSteps:', observedImagePath[0]);
+    }
+  });
 }

 module.exports = addImages;
--- a/api/app/clients/output_parsers/addImages.spec.js
+++ b/api/app/clients/output_parsers/addImages.spec.js
@@ -0,0 +1,142 @@
+let addImages = require('./addImages');
+
+describe('addImages', () => {
+  let intermediateSteps;
+  let responseMessage;
+  let options;
+
+  beforeEach(() => {
+    intermediateSteps = [];
+    responseMessage = { text: '' };
+    options = { debug: false };
+    this.options = options;
+    addImages = addImages.bind(this);
+  });
+
+  it('should handle null or undefined parameters', () => {
+    addImages(null, responseMessage);
+    expect(responseMessage.text).toBe('');
+
+    addImages(intermediateSteps, null);
+    expect(responseMessage.text).toBe('');
+
+    addImages(null, null);
+    expect(responseMessage.text).toBe('');
+  });
+
+  it('should append correct image markdown if not present in responseMessage', () => {
+    intermediateSteps.push({ observation: '![desc](/images/test.png)' });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('\n![desc](/images/test.png)');
+  });
+
+  it('should not append image markdown if already present in responseMessage', () => {
+    responseMessage.text = '![desc](/images/test.png)';
+    intermediateSteps.push({ observation: '![desc](/images/test.png)' });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('![desc](/images/test.png)');
+  });
+
+  it('should correct and append image markdown with erroneous URL', () => {
+    responseMessage.text = '![desc](sandbox:/images/test.png)';
+    intermediateSteps.push({ observation: '![desc](/images/test.png)' });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('![desc](/images/test.png)');
+  });
+
+  it('should correct multiple erroneous URLs in responseMessage', () => {
+    responseMessage.text =
+      '![desc1](sandbox:/images/test1.png) ![desc2](version:/images/test2.png)';
+    intermediateSteps.push({ observation: '![desc1](/images/test1.png)' });
+    intermediateSteps.push({ observation: '![desc2](/images/test2.png)' });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('![desc1](/images/test1.png) ![desc2](/images/test2.png)');
+  });
+
+  it('should not append non-image markdown observations', () => {
+    intermediateSteps.push({ observation: '[desc](/images/test.png)' });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('');
+  });
+
+  it('should handle multiple observations', () => {
+    intermediateSteps.push({ observation: '![desc1](/images/test1.png)' });
+    intermediateSteps.push({ observation: '![desc2](/images/test2.png)' });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('\n![desc1](/images/test1.png)\n![desc2](/images/test2.png)');
+  });
+
+  it('should not append if observation does not contain image markdown', () => {
+    intermediateSteps.push({ observation: 'This is a test observation without image markdown.' });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('');
+  });
+
+  it('should append correctly from a real scenario', () => {
+    responseMessage.text =
+      'Here is the generated image based on your request. It depicts a surreal landscape filled with floating musical notes. The style is impressionistic, with vibrant sunset hues dominating the scene. At the center, there\'s a silhouette of a grand piano, adding a dreamy emotion to the overall image. This could serve as a unique and creative music album cover. Would you like to make any changes or generate another image?';
+    const originalText = responseMessage.text;
+    const imageMarkdown = '![generated image](/images/img-RnVWaYo2Yg4x3e0isICiMuf5.png)';
+    intermediateSteps.push({ observation: imageMarkdown });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe(`${originalText}\n${imageMarkdown}`);
+  });
+
+  it('should extract only image markdowns when there is text between them', () => {
+    const markdownWithTextBetweenImages = `
+      ![image1](/images/image1.png)
+      Some text between images that should not be included.
+      ![image2](/images/image2.png)
+      More text that should be ignored.
+      ![image3](/images/image3.png)
+    `;
+    intermediateSteps.push({ observation: markdownWithTextBetweenImages });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('\n![image1](/images/image1.png)');
+  });
+
+  it('should only return the first image when multiple images are present', () => {
+    const markdownWithMultipleImages = `
+      ![image1](/images/image1.png)
+      ![image2](/images/image2.png)
+      ![image3](/images/image3.png)
+    `;
+    intermediateSteps.push({ observation: markdownWithMultipleImages });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('\n![image1](/images/image1.png)');
+  });
+
+  it('should not include any text or metadata surrounding the image markdown', () => {
+    const markdownWithMetadata = `
+      Title: Test Document
+      Author: John Doe
+      ![image1](/images/image1.png)
+      Some content after the image.
+      Vector values: [0.1, 0.2, 0.3]
+    `;
+    intermediateSteps.push({ observation: markdownWithMetadata });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('\n![image1](/images/image1.png)');
+  });
+
+  it('should handle complex markdown with multiple images and only return the first one', () => {
+    const complexMarkdown = `
+      # Document Title
+      
+      ## Section 1
+      Here's some text with an embedded image:
+      ![image1](/images/image1.png)
+      
+      ## Section 2
+      More text here...
+      ![image2](/images/image2.png)
+      
+      ### Subsection
+      Even more content
+      ![image3](/images/image3.png)
+    `;
+    intermediateSteps.push({ observation: complexMarkdown });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('\n![image1](/images/image1.png)');
+  });
+});
--- a/api/app/clients/output_parsers/index.js
+++ b/api/app/clients/output_parsers/index.js
@@ -1,9 +1,7 @@
 const addImages = require('./addImages');
-const handleInputs = require('./handleInputs');
 const handleOutputs = require('./handleOutputs');

 module.exports = {
  addImages,
-  ...handleInputs,
  ...handleOutputs,
 };
--- a/api/app/clients/prompts/addCacheControl.js
+++ b/api/app/clients/prompts/addCacheControl.js
@@ -0,0 +1,43 @@
+/**
+ * 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.
+ */
+function addCacheControl(messages) {
+  if (!Array.isArray(messages) || messages.length < 2) {
+    return messages;
+  }
+
+  const updatedMessages = [...messages];
+  let userMessagesModified = 0;
+
+  for (let i = updatedMessages.length - 1; i >= 0 && userMessagesModified < 2; i--) {
+    const message = updatedMessages[i];
+    if (message.role !== 'user') {
+      continue;
+    }
+
+    if (typeof message.content === 'string') {
+      message.content = [
+        {
+          type: 'text',
+          text: message.content,
+          cache_control: { type: 'ephemeral' },
+        },
+      ];
+      userMessagesModified++;
+    } else if (Array.isArray(message.content)) {
+      for (let j = message.content.length - 1; j >= 0; j--) {
+        if (message.content[j].type === 'text') {
+          message.content[j].cache_control = { type: 'ephemeral' };
+          userMessagesModified++;
+          break;
+        }
+      }
+    }
+  }
+
+  return updatedMessages;
+}
+
+module.exports = addCacheControl;
--- a/api/app/clients/prompts/addCacheControl.spec.js
+++ b/api/app/clients/prompts/addCacheControl.spec.js
@@ -0,0 +1,227 @@
+const addCacheControl = require('./addCacheControl');
+
+describe('addCacheControl', () => {
+  test('should add cache control to the last two user messages with array content', () => {
+    const messages = [
+      { role: 'user', content: [{ type: 'text', text: 'Hello' }] },
+      { role: 'assistant', content: [{ type: 'text', text: 'Hi there' }] },
+      { role: 'user', content: [{ type: 'text', text: 'How are you?' }] },
+      { role: 'assistant', content: [{ type: 'text', text: 'I\'m doing well, thanks!' }] },
+      { role: 'user', content: [{ type: 'text', text: 'Great!' }] },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content[0]).not.toHaveProperty('cache_control');
+    expect(result[2].content[0].cache_control).toEqual({ type: 'ephemeral' });
+    expect(result[4].content[0].cache_control).toEqual({ type: 'ephemeral' });
+  });
+
+  test('should add cache control to the last two user messages with string content', () => {
+    const messages = [
+      { role: 'user', content: 'Hello' },
+      { role: 'assistant', content: 'Hi there' },
+      { role: 'user', content: 'How are you?' },
+      { role: 'assistant', content: 'I\'m doing well, thanks!' },
+      { role: 'user', content: 'Great!' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content).toBe('Hello');
+    expect(result[2].content[0]).toEqual({
+      type: 'text',
+      text: 'How are you?',
+      cache_control: { type: 'ephemeral' },
+    });
+    expect(result[4].content[0]).toEqual({
+      type: 'text',
+      text: 'Great!',
+      cache_control: { type: 'ephemeral' },
+    });
+  });
+
+  test('should handle mixed string and array content', () => {
+    const messages = [
+      { role: 'user', content: 'Hello' },
+      { role: 'assistant', content: 'Hi there' },
+      { role: 'user', content: [{ type: 'text', text: 'How are you?' }] },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content[0]).toEqual({
+      type: 'text',
+      text: 'Hello',
+      cache_control: { type: 'ephemeral' },
+    });
+    expect(result[2].content[0].cache_control).toEqual({ type: 'ephemeral' });
+  });
+
+  test('should handle less than two user messages', () => {
+    const messages = [
+      { role: 'user', content: 'Hello' },
+      { role: 'assistant', content: 'Hi there' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content[0]).toEqual({
+      type: 'text',
+      text: 'Hello',
+      cache_control: { type: 'ephemeral' },
+    });
+    expect(result[1].content).toBe('Hi there');
+  });
+
+  test('should return original array if no user messages', () => {
+    const messages = [
+      { role: 'assistant', content: 'Hi there' },
+      { role: 'assistant', content: 'How can I help?' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result).toEqual(messages);
+  });
+
+  test('should handle empty array', () => {
+    const messages = [];
+    const result = addCacheControl(messages);
+    expect(result).toEqual([]);
+  });
+
+  test('should handle non-array input', () => {
+    const messages = 'not an array';
+    const result = addCacheControl(messages);
+    expect(result).toBe('not an array');
+  });
+
+  test('should not modify assistant messages', () => {
+    const messages = [
+      { role: 'user', content: 'Hello' },
+      { role: 'assistant', content: 'Hi there' },
+      { role: 'user', content: 'How are you?' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[1].content).toBe('Hi there');
+  });
+
+  test('should handle multiple content items in user messages', () => {
+    const messages = [
+      {
+        role: 'user',
+        content: [
+          { type: 'text', text: 'Hello' },
+          { type: 'image', url: 'http://example.com/image.jpg' },
+          { type: 'text', text: 'This is an image' },
+        ],
+      },
+      { role: 'assistant', content: 'Hi there' },
+      { role: 'user', content: 'How are you?' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content[0]).not.toHaveProperty('cache_control');
+    expect(result[0].content[1]).not.toHaveProperty('cache_control');
+    expect(result[0].content[2].cache_control).toEqual({ type: 'ephemeral' });
+    expect(result[2].content[0]).toEqual({
+      type: 'text',
+      text: 'How are you?',
+      cache_control: { type: 'ephemeral' },
+    });
+  });
+
+  test('should handle an array with mixed content types', () => {
+    const messages = [
+      { role: 'user', content: 'Hello' },
+      { role: 'assistant', content: 'Hi there' },
+      { role: 'user', content: [{ type: 'text', text: 'How are you?' }] },
+      { role: 'assistant', content: 'I\'m doing well, thanks!' },
+      { role: 'user', content: 'Great!' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content).toEqual('Hello');
+    expect(result[2].content[0]).toEqual({
+      type: 'text',
+      text: 'How are you?',
+      cache_control: { type: 'ephemeral' },
+    });
+    expect(result[4].content).toEqual([
+      {
+        type: 'text',
+        text: 'Great!',
+        cache_control: { type: 'ephemeral' },
+      },
+    ]);
+    expect(result[1].content).toBe('Hi there');
+    expect(result[3].content).toBe('I\'m doing well, thanks!');
+  });
+
+  test('should handle edge case with multiple content types', () => {
+    const messages = [
+      {
+        role: 'user',
+        content: [
+          {
+            type: 'image',
+            source: { type: 'base64', media_type: 'image/png', data: 'some_base64_string' },
+          },
+          {
+            type: 'image',
+            source: { type: 'base64', media_type: 'image/png', data: 'another_base64_string' },
+          },
+          { type: 'text', text: 'what do all these images have in common' },
+        ],
+      },
+      { role: 'assistant', content: 'I see multiple images.' },
+      { role: 'user', content: 'Correct!' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content[0]).not.toHaveProperty('cache_control');
+    expect(result[0].content[1]).not.toHaveProperty('cache_control');
+    expect(result[0].content[2].cache_control).toEqual({ type: 'ephemeral' });
+    expect(result[2].content[0]).toEqual({
+      type: 'text',
+      text: 'Correct!',
+      cache_control: { type: 'ephemeral' },
+    });
+  });
+
+  test('should handle user message with no text block', () => {
+    const messages = [
+      {
+        role: 'user',
+        content: [
+          {
+            type: 'image',
+            source: { type: 'base64', media_type: 'image/png', data: 'some_base64_string' },
+          },
+          {
+            type: 'image',
+            source: { type: 'base64', media_type: 'image/png', data: 'another_base64_string' },
+          },
+        ],
+      },
+      { role: 'assistant', content: 'I see two images.' },
+      { role: 'user', content: 'Correct!' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content[0]).not.toHaveProperty('cache_control');
+    expect(result[0].content[1]).not.toHaveProperty('cache_control');
+    expect(result[2].content[0]).toEqual({
+      type: 'text',
+      text: 'Correct!',
+      cache_control: { type: 'ephemeral' },
+    });
+  });
+});
--- a/api/app/clients/prompts/artifacts.js
+++ b/api/app/clients/prompts/artifacts.js
@@ -0,0 +1,527 @@
+const dedent = require('dedent');
+const { EModelEndpoint, ArtifactModes } = require('librechat-data-provider');
+const { generateShadcnPrompt } = require('~/app/clients/prompts/shadcn-docs/generate');
+const { components } = require('~/app/clients/prompts/shadcn-docs/components');
+
+// eslint-disable-next-line no-unused-vars
+const artifactsPromptV1 = dedent`The assistant can create and reference artifacts during conversations.
+  
+Artifacts are for substantial, self-contained content that users might modify or reuse, displayed in a separate UI window for clarity.
+
+# Good artifacts are...
+- Substantial content (>15 lines)
+- Content that the user is likely to modify, iterate on, or take ownership of
+- Self-contained, complex content that can be understood on its own, without context from the conversation
+- Content intended for eventual use outside the conversation (e.g., reports, emails, presentations)
+- Content likely to be referenced or reused multiple times
+
+# Don't use artifacts for...
+- Simple, informational, or short content, such as brief code snippets, mathematical equations, or small examples
+- Primarily explanatory, instructional, or illustrative content, such as examples provided to clarify a concept
+- Suggestions, commentary, or feedback on existing artifacts
+- Conversational or explanatory content that doesn't represent a standalone piece of work
+- Content that is dependent on the current conversational context to be useful
+- Content that is unlikely to be modified or iterated upon by the user
+- Request from users that appears to be a one-off question
+
+# Usage notes
+- One artifact per message unless specifically requested
+- Prefer in-line content (don't use artifacts) when possible. Unnecessary use of artifacts can be jarring for users.
+- If a user asks the assistant to "draw an SVG" or "make a website," the assistant does not need to explain that it doesn't have these capabilities. Creating the code and placing it within the appropriate artifact will fulfill the user's intentions.
+- If asked to generate an image, the assistant can offer an SVG instead. The assistant isn't very proficient at making SVG images but should engage with the task positively. Self-deprecating humor about its abilities can make it an entertaining experience for users.
+- The assistant errs on the side of simplicity and avoids overusing artifacts for content that can be effectively presented within the conversation.
+- Always provide complete, specific, and fully functional content without any placeholders, ellipses, or 'remains the same' comments.
+
+<artifact_instructions>
+  When collaborating with the user on creating content that falls into compatible categories, the assistant should follow these steps:
+
+  1. Create the artifact using the following format:
+
+     :::artifact{identifier="unique-identifier" type="mime-type" title="Artifact Title"}
+     \`\`\`
+     Your artifact content here
+     \`\`\`
+     :::
+
+  2. Assign an identifier to the \`identifier\` attribute. For updates, reuse the prior identifier. For new artifacts, the identifier should be descriptive and relevant to the content, using kebab-case (e.g., "example-code-snippet"). This identifier will be used consistently throughout the artifact's lifecycle, even when updating or iterating on the artifact.
+  3. Include a \`title\` attribute to provide a brief title or description of the content.
+  4. Add a \`type\` attribute to specify the type of content the artifact represents. Assign one of the following values to the \`type\` attribute:
+    - HTML: "text/html"
+      - The user interface can render single file HTML pages placed within the artifact tags. HTML, JS, and CSS should be in a single file when using the \`text/html\` type.
+      - Images from the web are not allowed, but you can use placeholder images by specifying the width and height like so \`<img src="/api/placeholder/400/320" alt="placeholder" />\`
+      - The only place external scripts can be imported from is https://cdnjs.cloudflare.com
+    - Mermaid Diagrams: "application/vnd.mermaid"
+      - The user interface will render Mermaid diagrams placed within the artifact tags.
+    - React Components: "application/vnd.react"
+      - Use this for displaying either: React elements, e.g. \`<strong>Hello World!</strong>\`, React pure functional components, e.g. \`() => <strong>Hello World!</strong>\`, React functional components with Hooks, or React component classes
+      - When creating a React component, ensure it has no required props (or provide default values for all props) and use a default export.
+      - Use Tailwind classes for styling. DO NOT USE ARBITRARY VALUES (e.g. \`h-[600px]\`).
+      - Base React is available to be imported. To use hooks, first import it at the top of the artifact, e.g. \`import { useState } from "react"\`
+      - The lucide-react@0.263.1 library is available to be imported. e.g. \`import { Camera } from "lucide-react"\` & \`<Camera color="red" size={48} />\`
+      - The recharts charting library is available to be imported, e.g. \`import { LineChart, XAxis, ... } from "recharts"\` & \`<LineChart ...><XAxis dataKey="name"> ...\`
+      - The assistant can use prebuilt components from the \`shadcn/ui\` library after it is imported: \`import { Alert, AlertDescription, AlertTitle, AlertDialog, AlertDialogAction } from '/components/ui/alert';\`. If using components from the shadcn/ui library, the assistant mentions this to the user and offers to help them install the components if necessary.
+      - Components MUST be imported from \`/components/ui/name\` and NOT from \`/components/name\` or \`@/components/ui/name\`.
+      - NO OTHER LIBRARIES (e.g. zod, hookform) ARE INSTALLED OR ABLE TO BE IMPORTED.
+      - Images from the web are not allowed, but you can use placeholder images by specifying the width and height like so \`<img src="/api/placeholder/400/320" alt="placeholder" />\`
+      - If you are unable to follow the above requirements for any reason, don't use artifacts and use regular code blocks instead, which will not attempt to render the component.
+  5. Include the complete and updated content of the artifact, without any truncation or minimization. Don't use "// rest of the code remains the same...".
+  6. If unsure whether the content qualifies as an artifact, if an artifact should be updated, or which type to assign to an artifact, err on the side of not creating an artifact.
+  7. Always use triple backticks (\`\`\`) to enclose the content within the artifact, regardless of the content type.
+</artifact_instructions>
+
+Here are some examples of correct usage of artifacts:
+
+<examples>
+  <example_docstring>
+    This example demonstrates how to create a Mermaid artifact for a simple flow chart.
+  </example_docstring>
+
+  <example>
+    <user_query>Can you create a simple flow chart showing the process of making tea using Mermaid?</user_query>
+
+    <assistant_response>
+      Sure! Here's a simple flow chart depicting the process of making tea using Mermaid syntax:
+
+      :::artifact{identifier="tea-making-flowchart" type="application/vnd.mermaid" title="Flow chart: Making Tea"}
+      \`\`\`mermaid
+      graph TD
+          A[Start] --> B{Water boiled?}
+          B -->|Yes| C[Add tea leaves to cup]
+          B -->|No| D[Boil water]
+          D --> B
+          C --> E[Pour boiling water into cup]
+          E --> F[Steep tea for desired time]
+          F --> G[Remove tea leaves]
+          G --> H[Add milk or sugar, if desired]
+          H --> I[Enjoy your tea!]
+          I --> J[End]
+      \`\`\`
+      :::
+
+      This flow chart uses Mermaid syntax to visualize the steps involved in making a cup of tea. Here's a brief explanation of the process:
+
+      1. Start
+      2. Check if water is boiled
+      3. If not boiled, boil the water
+      4. Once water is boiled, add tea leaves to the cup
+      5. Pour boiling water into the cup
+      6. Steep the tea for the desired time
+      7. Remove the tea leaves
+      8. Optionally add milk or sugar
+      9. Enjoy your tea!
+      10. End
+
+      This chart provides a clear visual representation of the tea-making process. You can easily modify or expand this chart if you want to add more details or steps to the process. Let me know if you'd like any changes or have any questions!
+    </assistant_response>
+  </example>
+</examples>`;
+const artifactsPrompt = dedent`The assistant can create and reference artifacts during conversations.
+  
+Artifacts are for substantial, self-contained content that users might modify or reuse, displayed in a separate UI window for clarity.
+
+# Good artifacts are...
+- Substantial content (>15 lines)
+- Content that the user is likely to modify, iterate on, or take ownership of
+- Self-contained, complex content that can be understood on its own, without context from the conversation
+- Content intended for eventual use outside the conversation (e.g., reports, emails, presentations)
+- Content likely to be referenced or reused multiple times
+
+# Don't use artifacts for...
+- Simple, informational, or short content, such as brief code snippets, mathematical equations, or small examples
+- Primarily explanatory, instructional, or illustrative content, such as examples provided to clarify a concept
+- Suggestions, commentary, or feedback on existing artifacts
+- Conversational or explanatory content that doesn't represent a standalone piece of work
+- Content that is dependent on the current conversational context to be useful
+- Content that is unlikely to be modified or iterated upon by the user
+- Request from users that appears to be a one-off question
+
+# Usage notes
+- One artifact per message unless specifically requested
+- Prefer in-line content (don't use artifacts) when possible. Unnecessary use of artifacts can be jarring for users.
+- If a user asks the assistant to "draw an SVG" or "make a website," the assistant does not need to explain that it doesn't have these capabilities. Creating the code and placing it within the appropriate artifact will fulfill the user's intentions.
+- If asked to generate an image, the assistant can offer an SVG instead. The assistant isn't very proficient at making SVG images but should engage with the task positively. Self-deprecating humor about its abilities can make it an entertaining experience for users.
+- The assistant errs on the side of simplicity and avoids overusing artifacts for content that can be effectively presented within the conversation.
+- Always provide complete, specific, and fully functional content for artifacts without any snippets, placeholders, ellipses, or 'remains the same' comments.
+- If an artifact is not necessary or requested, the assistant should not mention artifacts at all, and respond to the user accordingly.
+
+<artifact_instructions>
+  When collaborating with the user on creating content that falls into compatible categories, the assistant should follow these steps:
+
+  1. Create the artifact using the following format:
+
+     :::artifact{identifier="unique-identifier" type="mime-type" title="Artifact Title"}
+     \`\`\`
+     Your artifact content here
+     \`\`\`
+     :::
+
+  2. Assign an identifier to the \`identifier\` attribute. For updates, reuse the prior identifier. For new artifacts, the identifier should be descriptive and relevant to the content, using kebab-case (e.g., "example-code-snippet"). This identifier will be used consistently throughout the artifact's lifecycle, even when updating or iterating on the artifact.
+  3. Include a \`title\` attribute to provide a brief title or description of the content.
+  4. Add a \`type\` attribute to specify the type of content the artifact represents. Assign one of the following values to the \`type\` attribute:
+    - HTML: "text/html"
+      - The user interface can render single file HTML pages placed within the artifact tags. HTML, JS, and CSS should be in a single file when using the \`text/html\` type.
+      - Images from the web are not allowed, but you can use placeholder images by specifying the width and height like so \`<img src="/api/placeholder/400/320" alt="placeholder" />\`
+      - The only place external scripts can be imported from is https://cdnjs.cloudflare.com
+    - SVG: "image/svg+xml"
+      - The user interface will render the Scalable Vector Graphics (SVG) image within the artifact tags.
+      - The assistant should specify the viewbox of the SVG rather than defining a width/height
+    - Mermaid Diagrams: "application/vnd.mermaid"
+      - The user interface will render Mermaid diagrams placed within the artifact tags.
+    - React Components: "application/vnd.react"
+      - Use this for displaying either: React elements, e.g. \`<strong>Hello World!</strong>\`, React pure functional components, e.g. \`() => <strong>Hello World!</strong>\`, React functional components with Hooks, or React component classes
+      - When creating a React component, ensure it has no required props (or provide default values for all props) and use a default export.
+      - Use Tailwind classes for styling. DO NOT USE ARBITRARY VALUES (e.g. \`h-[600px]\`).
+      - Base React is available to be imported. To use hooks, first import it at the top of the artifact, e.g. \`import { useState } from "react"\`
+      - The lucide-react@0.394.0 library is available to be imported. e.g. \`import { Camera } from "lucide-react"\` & \`<Camera color="red" size={48} />\`
+      - The recharts charting library is available to be imported, e.g. \`import { LineChart, XAxis, ... } from "recharts"\` & \`<LineChart ...><XAxis dataKey="name"> ...\`
+      - The three.js library is available to be imported, e.g. \`import * as THREE from "three";\`
+      - The date-fns library is available to be imported, e.g. \`import { compareAsc, format } from "date-fns";\`
+      - The react-day-picker library is available to be imported, e.g. \`import { DayPicker } from "react-day-picker";\`
+      - The assistant can use prebuilt components from the \`shadcn/ui\` library after it is imported: \`import { Alert, AlertDescription, AlertTitle, AlertDialog, AlertDialogAction } from '/components/ui/alert';\`. If using components from the shadcn/ui library, the assistant mentions this to the user and offers to help them install the components if necessary.
+      - Components MUST be imported from \`/components/ui/name\` and NOT from \`/components/name\` or \`@/components/ui/name\`.
+      - NO OTHER LIBRARIES (e.g. zod, hookform) ARE INSTALLED OR ABLE TO BE IMPORTED.
+      - Images from the web are not allowed, but you can use placeholder images by specifying the width and height like so \`<img src="/api/placeholder/400/320" alt="placeholder" />\`
+      - When iterating on code, ensure that the code is complete and functional without any snippets, placeholders, or ellipses.
+      - If you are unable to follow the above requirements for any reason, don't use artifacts and use regular code blocks instead, which will not attempt to render the component.
+  5. Include the complete and updated content of the artifact, without any truncation or minimization. Don't use "// rest of the code remains the same...".
+  6. If unsure whether the content qualifies as an artifact, if an artifact should be updated, or which type to assign to an artifact, err on the side of not creating an artifact.
+  7. Always use triple backticks (\`\`\`) to enclose the content within the artifact, regardless of the content type.
+</artifact_instructions>
+
+Here are some examples of correct usage of artifacts:
+
+<examples>
+  <example_docstring>
+    This example demonstrates how to create a Mermaid artifact for a simple flow chart.
+  </example_docstring>
+
+  <example>
+    <user_query>Can you create a simple flow chart showing the process of making tea using Mermaid?</user_query>
+
+    <assistant_response>
+      Sure! Here's a simple flow chart depicting the process of making tea using Mermaid syntax:
+
+      :::artifact{identifier="tea-making-flowchart" type="application/vnd.mermaid" title="Flow chart: Making Tea"}
+      \`\`\`mermaid
+      graph TD
+          A[Start] --> B{Water boiled?}
+          B -->|Yes| C[Add tea leaves to cup]
+          B -->|No| D[Boil water]
+          D --> B
+          C --> E[Pour boiling water into cup]
+          E --> F[Steep tea for desired time]
+          F --> G[Remove tea leaves]
+          G --> H[Add milk or sugar, if desired]
+          H --> I[Enjoy your tea!]
+          I --> J[End]
+      \`\`\`
+      :::
+
+      This flow chart uses Mermaid syntax to visualize the steps involved in making a cup of tea. Here's a brief explanation of the process:
+
+      1. Start
+      2. Check if water is boiled
+      3. If not boiled, boil the water
+      4. Once water is boiled, add tea leaves to the cup
+      5. Pour boiling water into the cup
+      6. Steep the tea for the desired time
+      7. Remove the tea leaves
+      8. Optionally add milk or sugar
+      9. Enjoy your tea!
+      10. End
+
+      This chart provides a clear visual representation of the tea-making process. You can easily modify or expand this chart if you want to add more details or steps to the process. Let me know if you'd like any changes or have any questions!
+    </assistant_response>
+  </example>
+
+  <example>
+    <user_query>Create a simple React counter component</user_query>
+    <assistant_response>
+      Here's a simple React counter component:
+
+      :::artifact{identifier="react-counter" type="application/vnd.react" title="React Counter"}
+      \`\`\`
+      import { useState } from 'react';
+
+      export default function Counter() {
+        const [count, setCount] = useState(0);
+        return (
+          <div className="p-4">
+            <p className="mb-2">Count: {count}</p>
+            <button className="bg-blue-500 text-white px-4 py-2 rounded" onClick={() => setCount(count + 1)}>
+              Increment
+            </button>
+          </div>
+        );
+      }
+      \`\`\`
+      :::
+
+      This component creates a simple counter with an increment button.
+    </assistant_response>
+  </example>
+
+  <example>
+    <user_query>Create a basic HTML structure for a blog post</user_query>
+    <assistant_response>
+      Here's a basic HTML structure for a blog post:
+
+      :::artifact{identifier="blog-post-html" type="text/html" title="Blog Post HTML"}
+      \`\`\`
+      <!DOCTYPE html>
+      <html lang="en">
+      <head>
+        <meta charset="UTF-8">
+        <meta name="viewport" content="width=device-width, initial-scale=1.0">
+        <title>My Blog Post</title>
+        <style>
+          body { font-family: Arial, sans-serif; line-height: 1.6; max-width: 800px; margin: 0 auto; padding: 20px; }
+          h1 { color: #333; }
+          p { margin-bottom: 15px; }
+        </style>
+      </head>
+      <body>
+        <header>
+          <h1>My First Blog Post</h1>
+        </header>
+        <main>
+          <article>
+            <p>This is the content of my blog post. It's short and sweet!</p>
+          </article>
+        </main>
+        <footer>
+          <p>&copy; 2023 My Blog</p>
+        </footer>
+      </body>
+      </html>
+      \`\`\`
+      :::
+
+      This HTML structure provides a simple layout for a blog post.
+    </assistant_response>
+  </example>
+</examples>`;
+
+const artifactsOpenAIPrompt = dedent`The assistant can create and reference artifacts during conversations.
+  
+Artifacts are for substantial, self-contained content that users might modify or reuse, displayed in a separate UI window for clarity.
+
+# Good artifacts are...
+- Substantial content (>15 lines)
+- Content that the user is likely to modify, iterate on, or take ownership of
+- Self-contained, complex content that can be understood on its own, without context from the conversation
+- Content intended for eventual use outside the conversation (e.g., reports, emails, presentations)
+- Content likely to be referenced or reused multiple times
+
+# Don't use artifacts for...
+- Simple, informational, or short content, such as brief code snippets, mathematical equations, or small examples
+- Primarily explanatory, instructional, or illustrative content, such as examples provided to clarify a concept
+- Suggestions, commentary, or feedback on existing artifacts
+- Conversational or explanatory content that doesn't represent a standalone piece of work
+- Content that is dependent on the current conversational context to be useful
+- Content that is unlikely to be modified or iterated upon by the user
+- Request from users that appears to be a one-off question
+
+# Usage notes
+- One artifact per message unless specifically requested
+- Prefer in-line content (don't use artifacts) when possible. Unnecessary use of artifacts can be jarring for users.
+- If a user asks the assistant to "draw an SVG" or "make a website," the assistant does not need to explain that it doesn't have these capabilities. Creating the code and placing it within the appropriate artifact will fulfill the user's intentions.
+- If asked to generate an image, the assistant can offer an SVG instead. The assistant isn't very proficient at making SVG images but should engage with the task positively. Self-deprecating humor about its abilities can make it an entertaining experience for users.
+- The assistant errs on the side of simplicity and avoids overusing artifacts for content that can be effectively presented within the conversation.
+- Always provide complete, specific, and fully functional content for artifacts without any snippets, placeholders, ellipses, or 'remains the same' comments.
+- If an artifact is not necessary or requested, the assistant should not mention artifacts at all, and respond to the user accordingly.
+
+## Artifact Instructions
+  When collaborating with the user on creating content that falls into compatible categories, the assistant should follow these steps:
+
+  1. Create the artifact using the following remark-directive markdown format:
+
+      :::artifact{identifier="unique-identifier" type="mime-type" title="Artifact Title"}
+      \`\`\`
+      Your artifact content here
+      \`\`\`
+      :::
+
+  a. Example of correct format:
+
+      :::artifact{identifier="example-artifact" type="text/plain" title="Example Artifact"}
+      \`\`\`
+      This is the content of the artifact.
+      It can span multiple lines.
+      \`\`\`
+      :::
+
+  b. Common mistakes to avoid:
+   - Don't split the opening ::: line
+   - Don't add extra backticks outside the artifact structure
+   - Don't omit the closing :::
+
+  2. Assign an identifier to the \`identifier\` attribute. For updates, reuse the prior identifier. For new artifacts, the identifier should be descriptive and relevant to the content, using kebab-case (e.g., "example-code-snippet"). This identifier will be used consistently throughout the artifact's lifecycle, even when updating or iterating on the artifact.
+  3. Include a \`title\` attribute to provide a brief title or description of the content.
+  4. Add a \`type\` attribute to specify the type of content the artifact represents. Assign one of the following values to the \`type\` attribute:
+    - HTML: "text/html"
+      - The user interface can render single file HTML pages placed within the artifact tags. HTML, JS, and CSS should be in a single file when using the \`text/html\` type.
+      - Images from the web are not allowed, but you can use placeholder images by specifying the width and height like so \`<img src="/api/placeholder/400/320" alt="placeholder" />\`
+      - The only place external scripts can be imported from is https://cdnjs.cloudflare.com
+    - SVG: "image/svg+xml"
+      - The user interface will render the Scalable Vector Graphics (SVG) image within the artifact tags.
+      - The assistant should specify the viewbox of the SVG rather than defining a width/height
+    - Mermaid Diagrams: "application/vnd.mermaid"
+      - The user interface will render Mermaid diagrams placed within the artifact tags.
+    - React Components: "application/vnd.react"
+      - Use this for displaying either: React elements, e.g. \`<strong>Hello World!</strong>\`, React pure functional components, e.g. \`() => <strong>Hello World!</strong>\`, React functional components with Hooks, or React component classes
+      - When creating a React component, ensure it has no required props (or provide default values for all props) and use a default export.
+      - Use Tailwind classes for styling. DO NOT USE ARBITRARY VALUES (e.g. \`h-[600px]\`).
+      - Base React is available to be imported. To use hooks, first import it at the top of the artifact, e.g. \`import { useState } from "react"\`
+      - The lucide-react@0.394.0 library is available to be imported. e.g. \`import { Camera } from "lucide-react"\` & \`<Camera color="red" size={48} />\`
+      - The recharts charting library is available to be imported, e.g. \`import { LineChart, XAxis, ... } from "recharts"\` & \`<LineChart ...><XAxis dataKey="name"> ...\`
+      - The three.js library is available to be imported, e.g. \`import * as THREE from "three";\`
+      - The date-fns library is available to be imported, e.g. \`import { compareAsc, format } from "date-fns";\`
+      - The react-day-picker library is available to be imported, e.g. \`import { DayPicker } from "react-day-picker";\`
+      - The assistant can use prebuilt components from the \`shadcn/ui\` library after it is imported: \`import { Alert, AlertDescription, AlertTitle, AlertDialog, AlertDialogAction } from '/components/ui/alert';\`. If using components from the shadcn/ui library, the assistant mentions this to the user and offers to help them install the components if necessary.
+      - Components MUST be imported from \`/components/ui/name\` and NOT from \`/components/name\` or \`@/components/ui/name\`.
+      - NO OTHER LIBRARIES (e.g. zod, hookform) ARE INSTALLED OR ABLE TO BE IMPORTED.
+      - Images from the web are not allowed, but you can use placeholder images by specifying the width and height like so \`<img src="/api/placeholder/400/320" alt="placeholder" />\`
+      - When iterating on code, ensure that the code is complete and functional without any snippets, placeholders, or ellipses.
+      - If you are unable to follow the above requirements for any reason, don't use artifacts and use regular code blocks instead, which will not attempt to render the component.
+  5. Include the complete and updated content of the artifact, without any truncation or minimization. Don't use "// rest of the code remains the same...".
+  6. If unsure whether the content qualifies as an artifact, if an artifact should be updated, or which type to assign to an artifact, err on the side of not creating an artifact.
+  7. NEVER use triple backticks to enclose the artifact, ONLY the content within the artifact.
+
+Here are some examples of correct usage of artifacts:
+
+## Examples
+
+### Example 1
+
+    This example demonstrates how to create a Mermaid artifact for a simple flow chart.
+
+    User: Can you create a simple flow chart showing the process of making tea using Mermaid?
+
+    Assistant: Sure! Here's a simple flow chart depicting the process of making tea using Mermaid syntax:
+
+      :::artifact{identifier="tea-making-flowchart" type="application/vnd.mermaid" title="Flow chart: Making Tea"}
+      \`\`\`mermaid
+      graph TD
+          A[Start] --> B{Water boiled?}
+          B -->|Yes| C[Add tea leaves to cup]
+          B -->|No| D[Boil water]
+          D --> B
+          C --> E[Pour boiling water into cup]
+          E --> F[Steep tea for desired time]
+          F --> G[Remove tea leaves]
+          G --> H[Add milk or sugar, if desired]
+          H --> I[Enjoy your tea!]
+          I --> J[End]
+      \`\`\`
+      :::
+
+      This flow chart uses Mermaid syntax to visualize the steps involved in making a cup of tea. Here's a brief explanation of the process:
+
+      1. Start
+      2. Check if water is boiled
+      3. If not boiled, boil the water
+      4. Once water is boiled, add tea leaves to the cup
+      5. Pour boiling water into the cup
+      6. Steep the tea for the desired time
+      7. Remove the tea leaves
+      8. Optionally add milk or sugar
+      9. Enjoy your tea!
+      10. End
+
+      This chart provides a clear visual representation of the tea-making process. You can easily modify or expand this chart if you want to add more details or steps to the process. Let me know if you'd like any changes or have any questions!
+
+---
+
+### Example 2
+
+    User: Create a simple React counter component
+    
+    Assistant: Here's a simple React counter component:
+
+      :::artifact{identifier="react-counter" type="application/vnd.react" title="React Counter"}
+      \`\`\`
+      import { useState } from 'react';
+
+      export default function Counter() {
+        const [count, setCount] = useState(0);
+        return (
+          <div className="p-4">
+            <p className="mb-2">Count: {count}</p>
+            <button className="bg-blue-500 text-white px-4 py-2 rounded" onClick={() => setCount(count + 1)}>
+              Increment
+            </button>
+          </div>
+        );
+      }
+      \`\`\`
+      :::
+
+      This component creates a simple counter with an increment button.
+
+---
+
+### Example 3
+    User: Create a basic HTML structure for a blog post
+    Assistant: Here's a basic HTML structure for a blog post:
+
+      :::artifact{identifier="blog-post-html" type="text/html" title="Blog Post HTML"}
+      \`\`\`
+      <!DOCTYPE html>
+      <html lang="en">
+      <head>
+        <meta charset="UTF-8">
+        <meta name="viewport" content="width=device-width, initial-scale=1.0">
+        <title>My Blog Post</title>
+        <style>
+          body { font-family: Arial, sans-serif; line-height: 1.6; max-width: 800px; margin: 0 auto; padding: 20px; }
+          h1 { color: #333; }
+          p { margin-bottom: 15px; }
+        </style>
+      </head>
+      <body>
+        <header>
+          <h1>My First Blog Post</h1>
+        </header>
+        <main>
+          <article>
+            <p>This is the content of my blog post. It's short and sweet!</p>
+          </article>
+        </main>
+        <footer>
+          <p>&copy; 2023 My Blog</p>
+        </footer>
+      </body>
+      </html>
+      \`\`\`
+      :::
+
+      This HTML structure provides a simple layout for a blog post.
+
+---`;
+
+/**
+ *
+ * @param {Object} params
+ * @param {EModelEndpoint | string} params.endpoint - The current endpoint
+ * @param {ArtifactModes} params.artifacts - The current artifact mode
+ * @returns
+ */
+const generateArtifactsPrompt = ({ endpoint, artifacts }) => {
+  if (artifacts === ArtifactModes.CUSTOM) {
+    return null;
+  }
+
+  let prompt = artifactsPrompt;
+  if (endpoint !== EModelEndpoint.anthropic) {
+    prompt = artifactsOpenAIPrompt;
+  }
+
+  if (artifacts === ArtifactModes.SHADCNUI) {
+    prompt += generateShadcnPrompt({ components, useXML: endpoint === EModelEndpoint.anthropic });
+  }
+
+  return prompt;
+};
+
+module.exports = generateArtifactsPrompt;
--- a/api/app/clients/prompts/createContextHandlers.js
+++ b/api/app/clients/prompts/createContextHandlers.js
@@ -0,0 +1,160 @@
+const axios = require('axios');
+const { isEnabled } = require('~/server/utils');
+const { logger } = require('~/config');
+
+const footer = `Use the context as your learned knowledge to better answer the user.
+
+In your response, remember to follow these guidelines:
+- If you don't know the answer, simply say that you don't know.
+- If you are unsure how to answer, ask for clarification.
+- Avoid mentioning that you obtained the information from the context.
+`;
+
+function createContextHandlers(req, userMessageContent) {
+  if (!process.env.RAG_API_URL) {
+    return;
+  }
+
+  const queryPromises = [];
+  const processedFiles = [];
+  const processedIds = new Set();
+  const jwtToken = req.headers.authorization.split(' ')[1];
+  const useFullContext = isEnabled(process.env.RAG_USE_FULL_CONTEXT);
+
+  const query = async (file) => {
+    if (useFullContext) {
+      return axios.get(`${process.env.RAG_API_URL}/documents/${file.file_id}/context`, {
+        headers: {
+          Authorization: `Bearer ${jwtToken}`,
+        },
+      });
+    }
+
+    return axios.post(
+      `${process.env.RAG_API_URL}/query`,
+      {
+        file_id: file.file_id,
+        query: userMessageContent,
+        k: 4,
+      },
+      {
+        headers: {
+          Authorization: `Bearer ${jwtToken}`,
+          'Content-Type': 'application/json',
+        },
+      },
+    );
+  };
+
+  const processFile = async (file) => {
+    if (file.embedded && !processedIds.has(file.file_id)) {
+      try {
+        const promise = query(file);
+        queryPromises.push(promise);
+        processedFiles.push(file);
+        processedIds.add(file.file_id);
+      } catch (error) {
+        logger.error(`Error processing file ${file.filename}:`, error);
+      }
+    }
+  };
+
+  const createContext = async () => {
+    try {
+      if (!queryPromises.length || !processedFiles.length) {
+        return '';
+      }
+
+      const oneFile = processedFiles.length === 1;
+      const header = `The user has attached ${oneFile ? 'a' : processedFiles.length} file${
+        !oneFile ? 's' : ''
+      } to the conversation:`;
+
+      const files = `${
+        oneFile
+          ? ''
+          : `
+      <files>`
+      }${processedFiles
+        .map(
+          (file) => `
+              <file>
+                <filename>${file.filename}</filename>
+                <type>${file.type}</type>
+              </file>`,
+        )
+        .join('')}${
+        oneFile
+          ? ''
+          : `
+        </files>`
+      }`;
+
+      const resolvedQueries = await Promise.all(queryPromises);
+
+      const context =
+        resolvedQueries.length === 0
+          ? '\n\tThe semantic search did not return any results.'
+          : resolvedQueries
+            .map((queryResult, index) => {
+              const file = processedFiles[index];
+              let contextItems = queryResult.data;
+
+              const generateContext = (currentContext) =>
+                `
+          <file>
+            <filename>${file.filename}</filename>
+            <context>${currentContext}
+            </context>
+          </file>`;
+
+              if (useFullContext) {
+                return generateContext(`\n${contextItems}`);
+              }
+
+              contextItems = queryResult.data
+                .map((item) => {
+                  const pageContent = item[0].page_content;
+                  return `
+            <contextItem>
+              <![CDATA[${pageContent?.trim()}]]>
+            </contextItem>`;
+                })
+                .join('');
+
+              return generateContext(contextItems);
+            })
+            .join('');
+
+      if (useFullContext) {
+        const prompt = `${header}
+          ${context}
+          ${footer}`;
+
+        return prompt;
+      }
+
+      const prompt = `${header}
+        ${files}
+
+        A semantic search was executed with the user's message as the query, retrieving the following context inside <context></context> XML tags.
+
+        <context>${context}
+        </context>
+
+        ${footer}`;
+
+      return prompt;
+    } catch (error) {
+      logger.error('Error creating context:', error);
+      throw error;
+    }
+  };
+
+  return {
+    processFile,
+    createContext,
+  };
+}
+
+module.exports = createContextHandlers;
--- a/api/app/clients/prompts/createVisionPrompt.js
+++ b/api/app/clients/prompts/createVisionPrompt.js
@@ -0,0 +1,34 @@
+/**
+ * Generates a prompt instructing the user to describe an image in detail, tailored to different types of visual content.
+ * @param {boolean} pluralized - Whether to pluralize the prompt for multiple images.
+ * @returns {string} - The generated vision prompt.
+ */
+const createVisionPrompt = (pluralized = false) => {
+  return `Please describe the image${
+    pluralized ? 's' : ''
+  } in detail, covering relevant aspects such as:
+
+  For photographs, illustrations, or artwork:
+  - The main subject(s) and their appearance, positioning, and actions
+  - The setting, background, and any notable objects or elements
+  - Colors, lighting, and overall mood or atmosphere
+  - Any interesting details, textures, or patterns
+  - The style, technique, or medium used (if discernible)
+  
+  For screenshots or images containing text:
+  - The content and purpose of the text
+  - The layout, formatting, and organization of the information
+  - Any notable visual elements, such as logos, icons, or graphics
+  - The overall context or message conveyed by the screenshot
+  
+  For graphs, charts, or data visualizations:
+  - The type of graph or chart (e.g., bar graph, line chart, pie chart)
+  - The variables being compared or analyzed
+  - Any trends, patterns, or outliers in the data
+  - The axis labels, scales, and units of measurement
+  - The title, legend, and any additional context provided
+  
+  Be as specific and descriptive as possible while maintaining clarity and concision.`;
+};
+
+module.exports = createVisionPrompt;
--- a/api/app/clients/prompts/formatGoogleInputs.js
+++ b/api/app/clients/prompts/formatGoogleInputs.js
@@ -0,0 +1,42 @@
+/**
+ * Formats an object to match the struct_val, list_val, string_val, float_val, and int_val format.
+ *
+ * @param {Object} obj - The object to be formatted.
+ * @returns {Object} The formatted object.
+ *
+ * Handles different types:
+ * - Arrays are wrapped in list_val and each element is processed.
+ * - Objects are recursively processed.
+ * - Strings are wrapped in string_val.
+ * - Numbers are wrapped in float_val or int_val depending on whether they are floating-point or integers.
+ */
+function formatGoogleInputs(obj) {
+  const formattedObj = {};
+
+  for (const key in obj) {
+    if (Object.prototype.hasOwnProperty.call(obj, key)) {
+      const value = obj[key];
+
+      // Handle arrays
+      if (Array.isArray(value)) {
+        formattedObj[key] = { list_val: value.map((item) => formatGoogleInputs(item)) };
+      }
+      // Handle objects
+      else if (typeof value === 'object' && value !== null) {
+        formattedObj[key] = formatGoogleInputs(value);
+      }
+      // Handle numbers
+      else if (typeof value === 'number') {
+        formattedObj[key] = Number.isInteger(value) ? { int_val: value } : { float_val: value };
+      }
+      // Handle other types (e.g., strings)
+      else {
+        formattedObj[key] = { string_val: [value] };
+      }
+    }
+  }
+
+  return { struct_val: formattedObj };
+}
+
+module.exports = formatGoogleInputs;
--- a/api/app/clients/prompts/formatGoogleInputs.spec.js
+++ b/api/app/clients/prompts/formatGoogleInputs.spec.js
@@ -0,0 +1,274 @@
+const formatGoogleInputs = require('./formatGoogleInputs');
+
+describe('formatGoogleInputs', () => {
+  it('formats message correctly', () => {
+    const input = {
+      messages: [
+        {
+          content: 'hi',
+          author: 'user',
+        },
+      ],
+      context: 'context',
+      examples: [
+        {
+          input: {
+            author: 'user',
+            content: 'user input',
+          },
+          output: {
+            author: 'bot',
+            content: 'bot output',
+          },
+        },
+      ],
+      parameters: {
+        temperature: 0.2,
+        topP: 0.8,
+        topK: 40,
+        maxOutputTokens: 1024,
+      },
+    };
+
+    const expectedOutput = {
+      struct_val: {
+        messages: {
+          list_val: [
+            {
+              struct_val: {
+                content: {
+                  string_val: ['hi'],
+                },
+                author: {
+                  string_val: ['user'],
+                },
+              },
+            },
+          ],
+        },
+        context: {
+          string_val: ['context'],
+        },
+        examples: {
+          list_val: [
+            {
+              struct_val: {
+                input: {
+                  struct_val: {
+                    author: {
+                      string_val: ['user'],
+                    },
+                    content: {
+                      string_val: ['user input'],
+                    },
+                  },
+                },
+                output: {
+                  struct_val: {
+                    author: {
+                      string_val: ['bot'],
+                    },
+                    content: {
+                      string_val: ['bot output'],
+                    },
+                  },
+                },
+              },
+            },
+          ],
+        },
+        parameters: {
+          struct_val: {
+            temperature: {
+              float_val: 0.2,
+            },
+            topP: {
+              float_val: 0.8,
+            },
+            topK: {
+              int_val: 40,
+            },
+            maxOutputTokens: {
+              int_val: 1024,
+            },
+          },
+        },
+      },
+    };
+
+    const result = formatGoogleInputs(input);
+    expect(JSON.stringify(result)).toEqual(JSON.stringify(expectedOutput));
+  });
+
+  it('formats real payload parts', () => {
+    const input = {
+      instances: [
+        {
+          context: 'context',
+          examples: [
+            {
+              input: {
+                author: 'user',
+                content: 'user input',
+              },
+              output: {
+                author: 'bot',
+                content: 'user output',
+              },
+            },
+          ],
+          messages: [
+            {
+              author: 'user',
+              content: 'hi',
+            },
+          ],
+        },
+      ],
+      parameters: {
+        candidateCount: 1,
+        maxOutputTokens: 1024,
+        temperature: 0.2,
+        topP: 0.8,
+        topK: 40,
+      },
+    };
+    const expectedOutput = {
+      struct_val: {
+        instances: {
+          list_val: [
+            {
+              struct_val: {
+                context: { string_val: ['context'] },
+                examples: {
+                  list_val: [
+                    {
+                      struct_val: {
+                        input: {
+                          struct_val: {
+                            author: { string_val: ['user'] },
+                            content: { string_val: ['user input'] },
+                          },
+                        },
+                        output: {
+                          struct_val: {
+                            author: { string_val: ['bot'] },
+                            content: { string_val: ['user output'] },
+                          },
+                        },
+                      },
+                    },
+                  ],
+                },
+                messages: {
+                  list_val: [
+                    {
+                      struct_val: {
+                        author: { string_val: ['user'] },
+                        content: { string_val: ['hi'] },
+                      },
+                    },
+                  ],
+                },
+              },
+            },
+          ],
+        },
+        parameters: {
+          struct_val: {
+            candidateCount: { int_val: 1 },
+            maxOutputTokens: { int_val: 1024 },
+            temperature: { float_val: 0.2 },
+            topP: { float_val: 0.8 },
+            topK: { int_val: 40 },
+          },
+        },
+      },
+    };
+
+    const result = formatGoogleInputs(input);
+    expect(JSON.stringify(result)).toEqual(JSON.stringify(expectedOutput));
+  });
+
+  it('helps create valid payload parts', () => {
+    const instances = {
+      context: 'context',
+      examples: [
+        {
+          input: {
+            author: 'user',
+            content: 'user input',
+          },
+          output: {
+            author: 'bot',
+            content: 'user output',
+          },
+        },
+      ],
+      messages: [
+        {
+          author: 'user',
+          content: 'hi',
+        },
+      ],
+    };
+
+    const expectedInstances = {
+      struct_val: {
+        context: { string_val: ['context'] },
+        examples: {
+          list_val: [
+            {
+              struct_val: {
+                input: {
+                  struct_val: {
+                    author: { string_val: ['user'] },
+                    content: { string_val: ['user input'] },
+                  },
+                },
+                output: {
+                  struct_val: {
+                    author: { string_val: ['bot'] },
+                    content: { string_val: ['user output'] },
+                  },
+                },
+              },
+            },
+          ],
+        },
+        messages: {
+          list_val: [
+            {
+              struct_val: {
+                author: { string_val: ['user'] },
+                content: { string_val: ['hi'] },
+              },
+            },
+          ],
+        },
+      },
+    };
+
+    const parameters = {
+      candidateCount: 1,
+      maxOutputTokens: 1024,
+      temperature: 0.2,
+      topP: 0.8,
+      topK: 40,
+    };
+    const expectedParameters = {
+      struct_val: {
+        candidateCount: { int_val: 1 },
+        maxOutputTokens: { int_val: 1024 },
+        temperature: { float_val: 0.2 },
+        topP: { float_val: 0.8 },
+        topK: { int_val: 40 },
+      },
+    };
+
+    const instancesResult = formatGoogleInputs(instances);
+    const parametersResult = formatGoogleInputs(parameters);
+    expect(JSON.stringify(instancesResult)).toEqual(JSON.stringify(expectedInstances));
+    expect(JSON.stringify(parametersResult)).toEqual(JSON.stringify(expectedParameters));
+  });
+});
--- a/api/app/clients/prompts/formatMessages.js
+++ b/api/app/clients/prompts/formatMessages.js
@@ -0,0 +1,134 @@
+const { EModelEndpoint } = require('librechat-data-provider');
+const { HumanMessage, AIMessage, SystemMessage } = require('langchain/schema');
+
+/**
+ * Formats a message to OpenAI Vision API payload format.
+ *
+ * @param {Object} params - The parameters for formatting.
+ * @param {Object} params.message - The message object to format.
+ * @param {string} [params.message.role] - The role of the message sender (must be 'user').
+ * @param {string} [params.message.content] - The text content of the message.
+ * @param {EModelEndpoint} [params.endpoint] - Identifier for specific endpoint handling
+ * @param {Array<string>} [params.image_urls] - The image_urls to attach to the message.
+ * @returns {(Object)} - The formatted message.
+ */
+const formatVisionMessage = ({ message, image_urls, endpoint }) => {
+  if (endpoint === EModelEndpoint.anthropic) {
+    message.content = [...image_urls, { type: 'text', text: message.content }];
+    return message;
+  }
+
+  message.content = [{ type: 'text', text: message.content }, ...image_urls];
+
+  return message;
+};
+
+/**
+ * Formats a message to OpenAI payload format based on the provided options.
+ *
+ * @param {Object} params - The parameters for formatting.
+ * @param {Object} params.message - The message object to format.
+ * @param {string} [params.message.role] - The role of the message sender (e.g., 'user', 'assistant').
+ * @param {string} [params.message._name] - The name associated with the message.
+ * @param {string} [params.message.sender] - The sender of the message.
+ * @param {string} [params.message.text] - The text content of the message.
+ * @param {string} [params.message.content] - The content of the message.
+ * @param {Array<string>} [params.message.image_urls] - The image_urls attached to the message for Vision API.
+ * @param {string} [params.userName] - The name of the user.
+ * @param {string} [params.assistantName] - The name of the assistant.
+ * @param {string} [params.endpoint] - Identifier for specific endpoint handling
+ * @param {boolean} [params.langChain=false] - Whether to return a LangChain message object.
+ * @returns {(Object|HumanMessage|AIMessage|SystemMessage)} - The formatted message.
+ */
+const formatMessage = ({ message, userName, assistantName, endpoint, langChain = false }) => {
+  let { role: _role, _name, sender, text, content: _content, lc_id } = message;
+  if (lc_id && lc_id[2] && !langChain) {
+    const roleMapping = {
+      SystemMessage: 'system',
+      HumanMessage: 'user',
+      AIMessage: 'assistant',
+    };
+    _role = roleMapping[lc_id[2]];
+  }
+  const role = _role ?? (sender && sender?.toLowerCase() === 'user' ? 'user' : 'assistant');
+  const content = text ?? _content ?? '';
+  const formattedMessage = {
+    role,
+    content,
+  };
+
+  const { image_urls } = message;
+  if (Array.isArray(image_urls) && image_urls.length > 0 && role === 'user') {
+    return formatVisionMessage({
+      message: formattedMessage,
+      image_urls: message.image_urls,
+      endpoint,
+    });
+  }
+
+  if (_name) {
+    formattedMessage.name = _name;
+  }
+
+  if (userName && formattedMessage.role === 'user') {
+    formattedMessage.name = userName;
+  }
+
+  if (assistantName && formattedMessage.role === 'assistant') {
+    formattedMessage.name = assistantName;
+  }
+
+  if (formattedMessage.name) {
+    // Conform to API regex: ^[a-zA-Z0-9_-]{1,64}$
+    // https://community.openai.com/t/the-format-of-the-name-field-in-the-documentation-is-incorrect/175684/2
+    formattedMessage.name = formattedMessage.name.replace(/[^a-zA-Z0-9_-]/g, '_');
+
+    if (formattedMessage.name.length > 64) {
+      formattedMessage.name = formattedMessage.name.substring(0, 64);
+    }
+  }
+
+  if (!langChain) {
+    return formattedMessage;
+  }
+
+  if (role === 'user') {
+    return new HumanMessage(formattedMessage);
+  } else if (role === 'assistant') {
+    return new AIMessage(formattedMessage);
+  } else {
+    return new SystemMessage(formattedMessage);
+  }
+};
+
+/**
+ * Formats an array of messages for LangChain.
+ *
+ * @param {Array<Object>} messages - The array of messages to format.
+ * @param {Object} formatOptions - The options for formatting each message.
+ * @param {string} [formatOptions.userName] - The name of the user.
+ * @param {string} [formatOptions.assistantName] - The name of the assistant.
+ * @returns {Array<(HumanMessage|AIMessage|SystemMessage)>} - The array of formatted LangChain messages.
+ */
+const formatLangChainMessages = (messages, formatOptions) =>
+  messages.map((msg) => formatMessage({ ...formatOptions, message: msg, langChain: true }));
+
+/**
+ * Formats a LangChain message object by merging properties from `lc_kwargs` or `kwargs` and `additional_kwargs`.
+ *
+ * @param {Object} message - The message object to format.
+ * @param {Object} [message.lc_kwargs] - Contains properties to be merged. Either this or `message.kwargs` should be provided.
+ * @param {Object} [message.kwargs] - Contains properties to be merged. Either this or `message.lc_kwargs` should be provided.
+ * @param {Object} [message.kwargs.additional_kwargs] - Additional properties to be merged.
+ *
+ * @returns {Object} The formatted LangChain message.
+ */
+const formatFromLangChain = (message) => {
+  const { additional_kwargs, ...message_kwargs } = message.lc_kwargs ?? message.kwargs;
+  return {
+    ...message_kwargs,
+    ...additional_kwargs,
+  };
+};
+
+module.exports = { formatMessage, formatLangChainMessages, formatFromLangChain };
--- a/api/app/clients/prompts/formatMessages.spec.js
+++ b/api/app/clients/prompts/formatMessages.spec.js
@@ -0,0 +1,277 @@
+const { Constants } = require('librechat-data-provider');
+const { HumanMessage, AIMessage, SystemMessage } = require('langchain/schema');
+const { formatMessage, formatLangChainMessages, formatFromLangChain } = require('./formatMessages');
+
+describe('formatMessage', () => {
+  it('formats user message', () => {
+    const input = {
+      message: {
+        sender: 'user',
+        text: 'Hello',
+      },
+      userName: 'John',
+    };
+    const result = formatMessage(input);
+    expect(result).toEqual({
+      role: 'user',
+      content: 'Hello',
+      name: 'John',
+    });
+  });
+
+  it('sanitizes the name by replacing invalid characters (per OpenAI)', () => {
+    const input = {
+      message: {
+        sender: 'user',
+        text: 'Hello',
+      },
+      userName: ' John$Doe@Example! ',
+    };
+    const result = formatMessage(input);
+    expect(result).toEqual({
+      role: 'user',
+      content: 'Hello',
+      name: '_John_Doe_Example__',
+    });
+  });
+
+  it('trims the name to a maximum length of 64 characters', () => {
+    const longName = 'a'.repeat(100);
+    const input = {
+      message: {
+        sender: 'user',
+        text: 'Hello',
+      },
+      userName: longName,
+    };
+    const result = formatMessage(input);
+    expect(result.name.length).toBe(64);
+    expect(result.name).toBe('a'.repeat(64));
+  });
+
+  it('formats a realistic user message', () => {
+    const input = {
+      message: {
+        _id: '6512cdfb92cbf69fea615331',
+        messageId: 'b620bf73-c5c3-4a38-b724-76886aac24c4',
+        __v: 0,
+        conversationId: '5c23d24f-941f-4aab-85df-127b596c8aa5',
+        createdAt: Date.now(),
+        error: false,
+        finish_reason: null,
+        isCreatedByUser: true,
+        isEdited: false,
+        model: null,
+        parentMessageId: Constants.NO_PARENT,
+        sender: 'User',
+        text: 'hi',
+        tokenCount: 5,
+        unfinished: false,
+        updatedAt: Date.now(),
+        user: '6512cdf475f05c86d44c31d2',
+      },
+      userName: 'John',
+    };
+    const result = formatMessage(input);
+    expect(result).toEqual({
+      role: 'user',
+      content: 'hi',
+      name: 'John',
+    });
+  });
+
+  it('formats assistant message', () => {
+    const input = {
+      message: {
+        sender: 'assistant',
+        text: 'Hi there',
+      },
+      assistantName: 'Assistant',
+    };
+    const result = formatMessage(input);
+    expect(result).toEqual({
+      role: 'assistant',
+      content: 'Hi there',
+      name: 'Assistant',
+    });
+  });
+
+  it('formats system message', () => {
+    const input = {
+      message: {
+        role: 'system',
+        text: 'Hi there',
+      },
+    };
+    const result = formatMessage(input);
+    expect(result).toEqual({
+      role: 'system',
+      content: 'Hi there',
+    });
+  });
+
+  it('formats user message with langChain', () => {
+    const input = {
+      message: {
+        sender: 'user',
+        text: 'Hello',
+      },
+      userName: 'John',
+      langChain: true,
+    };
+    const result = formatMessage(input);
+    expect(result).toBeInstanceOf(HumanMessage);
+    expect(result.lc_kwargs.content).toEqual(input.message.text);
+    expect(result.lc_kwargs.name).toEqual(input.userName);
+  });
+
+  it('formats assistant message with langChain', () => {
+    const input = {
+      message: {
+        sender: 'assistant',
+        text: 'Hi there',
+      },
+      assistantName: 'Assistant',
+      langChain: true,
+    };
+    const result = formatMessage(input);
+    expect(result).toBeInstanceOf(AIMessage);
+    expect(result.lc_kwargs.content).toEqual(input.message.text);
+    expect(result.lc_kwargs.name).toEqual(input.assistantName);
+  });
+
+  it('formats system message with langChain', () => {
+    const input = {
+      message: {
+        role: 'system',
+        text: 'This is a system message.',
+      },
+      langChain: true,
+    };
+    const result = formatMessage(input);
+    expect(result).toBeInstanceOf(SystemMessage);
+    expect(result.lc_kwargs.content).toEqual(input.message.text);
+  });
+
+  it('formats langChain messages into OpenAI payload format', () => {
+    const human = {
+      message: new HumanMessage({
+        content: 'Hello',
+      }),
+    };
+    const system = {
+      message: new SystemMessage({
+        content: 'Hello',
+      }),
+    };
+    const ai = {
+      message: new AIMessage({
+        content: 'Hello',
+      }),
+    };
+    const humanResult = formatMessage(human);
+    const systemResult = formatMessage(system);
+    const aiResult = formatMessage(ai);
+    expect(humanResult).toEqual({
+      role: 'user',
+      content: 'Hello',
+    });
+    expect(systemResult).toEqual({
+      role: 'system',
+      content: 'Hello',
+    });
+    expect(aiResult).toEqual({
+      role: 'assistant',
+      content: 'Hello',
+    });
+  });
+});
+
+describe('formatLangChainMessages', () => {
+  it('formats an array of messages for LangChain', () => {
+    const messages = [
+      {
+        role: 'system',
+        content: 'This is a system message',
+      },
+      {
+        sender: 'user',
+        text: 'Hello',
+      },
+      {
+        sender: 'assistant',
+        text: 'Hi there',
+      },
+    ];
+    const formatOptions = {
+      userName: 'John',
+      assistantName: 'Assistant',
+    };
+    const result = formatLangChainMessages(messages, formatOptions);
+    expect(result).toHaveLength(3);
+    expect(result[0]).toBeInstanceOf(SystemMessage);
+    expect(result[1]).toBeInstanceOf(HumanMessage);
+    expect(result[2]).toBeInstanceOf(AIMessage);
+
+    expect(result[0].lc_kwargs.content).toEqual(messages[0].content);
+    expect(result[1].lc_kwargs.content).toEqual(messages[1].text);
+    expect(result[2].lc_kwargs.content).toEqual(messages[2].text);
+
+    expect(result[1].lc_kwargs.name).toEqual(formatOptions.userName);
+    expect(result[2].lc_kwargs.name).toEqual(formatOptions.assistantName);
+  });
+
+  describe('formatFromLangChain', () => {
+    it('should merge kwargs and additional_kwargs', () => {
+      const message = {
+        kwargs: {
+          content: 'some content',
+          name: 'dan',
+          additional_kwargs: {
+            function_call: {
+              name: 'dall-e',
+              arguments: '{\n  "input": "Subject: hedgehog, Style: cute"\n}',
+            },
+          },
+        },
+      };
+
+      const expected = {
+        content: 'some content',
+        name: 'dan',
+        function_call: {
+          name: 'dall-e',
+          arguments: '{\n  "input": "Subject: hedgehog, Style: cute"\n}',
+        },
+      };
+
+      expect(formatFromLangChain(message)).toEqual(expected);
+    });
+
+    it('should handle messages without additional_kwargs', () => {
+      const message = {
+        kwargs: {
+          content: 'some content',
+          name: 'dan',
+        },
+      };
+
+      const expected = {
+        content: 'some content',
+        name: 'dan',
+      };
+
+      expect(formatFromLangChain(message)).toEqual(expected);
+    });
+
+    it('should handle empty messages', () => {
+      const message = {
+        kwargs: {},
+      };
+
+      const expected = {};
+
+      expect(formatFromLangChain(message)).toEqual(expected);
+    });
+  });
+});
--- a/api/app/clients/output_parsers/handleInputs.js
+++ b/api/app/clients/output_parsers/handleInputs.js
--- a/api/app/clients/prompts/index.js
+++ b/api/app/clients/prompts/index.js
@@ -1,9 +1,21 @@
+const addCacheControl = require('./addCacheControl');
+const formatMessages = require('./formatMessages');
+const summaryPrompts = require('./summaryPrompts');
+const handleInputs = require('./handleInputs');
 const instructions = require('./instructions');
 const titlePrompts = require('./titlePrompts');
-const refinePrompts = require('./refinePrompts');
+const truncateText = require('./truncateText');
+const createVisionPrompt = require('./createVisionPrompt');
+const createContextHandlers = require('./createContextHandlers');

 module.exports = {
-  ...refinePrompts,
+  addCacheControl,
+  ...formatMessages,
+  ...summaryPrompts,
+  ...handleInputs,
  ...instructions,
  ...titlePrompts,
+  ...truncateText,
+  createVisionPrompt,
+  createContextHandlers,
 };
--- a/api/app/clients/prompts/refinePrompts.js
+++ b/api/app/clients/prompts/refinePrompts.js
@@ -1,24 +0,0 @@
-const { PromptTemplate } = require('langchain/prompts');
-
-const refinePromptTemplate = `Your job is to produce a final summary of the following conversation.
-We have provided an existing summary up to a certain point: "{existing_answer}"
-We have the opportunity to refine the existing summary
-(only if needed) with some more context below.
------------
-"{text}"
------------
-
-Given the new context, refine the original summary of the conversation.
-Do note who is speaking in the conversation to give proper context.
-If the context isn't useful, return the original summary.
-
-REFINED CONVERSATION SUMMARY:`;
-
-const refinePrompt = new PromptTemplate({
-  template: refinePromptTemplate,
-  inputVariables: ['existing_answer', 'text'],
-});
-
-module.exports = {
-  refinePrompt,
-};
--- a/api/app/clients/prompts/shadcn-docs/components.js
+++ b/api/app/clients/prompts/shadcn-docs/components.js
@@ -0,0 +1,495 @@
+// Essential Components
+const essentialComponents = {
+  avatar: {
+    componentName: 'Avatar',
+    importDocs: 'import { Avatar, AvatarFallback, AvatarImage } from "/components/ui/avatar"',
+    usageDocs: `
+<Avatar>
+  <AvatarImage src="https://github.com/shadcn.png" />
+  <AvatarFallback>CN</AvatarFallback>
+</Avatar>`,
+  },
+  button: {
+    componentName: 'Button',
+    importDocs: 'import { Button } from "/components/ui/button"',
+    usageDocs: `
+<Button variant="outline">Button</Button>`,
+  },
+  card: {
+    componentName: 'Card',
+    importDocs: `
+import {
+  Card,
+  CardContent,
+  CardDescription,
+  CardFooter,
+  CardHeader,
+  CardTitle,
+} from "/components/ui/card"`,
+    usageDocs: `
+<Card>
+  <CardHeader>
+    <CardTitle>Card Title</CardTitle>
+    <CardDescription>Card Description</CardDescription>
+  </CardHeader>
+  <CardContent>
+    <p>Card Content</p>
+  </CardContent>
+  <CardFooter>
+    <p>Card Footer</p>
+  </CardFooter>
+</Card>`,
+  },
+  checkbox: {
+    componentName: 'Checkbox',
+    importDocs: 'import { Checkbox } from "/components/ui/checkbox"',
+    usageDocs: '<Checkbox />',
+  },
+  input: {
+    componentName: 'Input',
+    importDocs: 'import { Input } from "/components/ui/input"',
+    usageDocs: '<Input />',
+  },
+  label: {
+    componentName: 'Label',
+    importDocs: 'import { Label } from "/components/ui/label"',
+    usageDocs: '<Label htmlFor="email">Your email address</Label>',
+  },
+  radioGroup: {
+    componentName: 'RadioGroup',
+    importDocs: `
+import { Label } from "/components/ui/label"
+import { RadioGroup, RadioGroupItem } from "/components/ui/radio-group"`,
+    usageDocs: `
+<RadioGroup defaultValue="option-one">
+  <div className="flex items-center space-x-2">
+    <RadioGroupItem value="option-one" id="option-one" />
+    <Label htmlFor="option-one">Option One</Label>
+  </div>
+  <div className="flex items-center space-x-2">
+    <RadioGroupItem value="option-two" id="option-two" />
+    <Label htmlFor="option-two">Option Two</Label>
+  </div>
+</RadioGroup>`,
+  },
+  select: {
+    componentName: 'Select',
+    importDocs: `
+import {
+  Select,
+  SelectContent,
+  SelectItem,
+  SelectTrigger,
+  SelectValue,
+} from "/components/ui/select"`,
+    usageDocs: `
+<Select>
+  <SelectTrigger className="w-[180px]">
+    <SelectValue placeholder="Theme" />
+  </SelectTrigger>
+  <SelectContent>
+    <SelectItem value="light">Light</SelectItem>
+    <SelectItem value="dark">Dark</SelectItem>
+    <SelectItem value="system">System</SelectItem>
+  </SelectContent>
+</Select>`,
+  },
+  textarea: {
+    componentName: 'Textarea',
+    importDocs: 'import { Textarea } from "/components/ui/textarea"',
+    usageDocs: '<Textarea />',
+  },
+};
+
+// Extra Components
+const extraComponents = {
+  accordion: {
+    componentName: 'Accordion',
+    importDocs: `
+import {
+  Accordion,
+  AccordionContent,
+  AccordionItem,
+  AccordionTrigger,
+} from "/components/ui/accordion"`,
+    usageDocs: `
+<Accordion type="single" collapsible>
+  <AccordionItem value="item-1">
+    <AccordionTrigger>Is it accessible?</AccordionTrigger>
+    <AccordionContent>
+      Yes. It adheres to the WAI-ARIA design pattern.
+    </AccordionContent>
+  </AccordionItem>
+</Accordion>`,
+  },
+  alertDialog: {
+    componentName: 'AlertDialog',
+    importDocs: `
+import {
+  AlertDialog,
+  AlertDialogAction,
+  AlertDialogCancel,
+  AlertDialogContent,
+  AlertDialogDescription,
+  AlertDialogFooter,
+  AlertDialogHeader,
+  AlertDialogTitle,
+  AlertDialogTrigger,
+} from "/components/ui/alert-dialog"`,
+    usageDocs: `
+<AlertDialog>
+  <AlertDialogTrigger>Open</AlertDialogTrigger>
+  <AlertDialogContent>
+    <AlertDialogHeader>
+      <AlertDialogTitle>Are you absolutely sure?</AlertDialogTitle>
+      <AlertDialogDescription>
+        This action cannot be undone.
+      </AlertDialogDescription>
+    </AlertDialogHeader>
+    <AlertDialogFooter>
+      <AlertDialogCancel>Cancel</AlertDialogCancel>
+      <AlertDialogAction>Continue</AlertDialogAction>
+    </AlertDialogFooter>
+  </AlertDialogContent>
+</AlertDialog>`,
+  },
+  alert: {
+    componentName: 'Alert',
+    importDocs: `
+import {
+  Alert,
+  AlertDescription,
+  AlertTitle,
+} from "/components/ui/alert"`,
+    usageDocs: `
+<Alert>
+  <AlertTitle>Heads up!</AlertTitle>
+  <AlertDescription>
+    You can add components to your app using the cli.
+  </AlertDescription>
+</Alert>`,
+  },
+  aspectRatio: {
+    componentName: 'AspectRatio',
+    importDocs: 'import { AspectRatio } from "/components/ui/aspect-ratio"',
+    usageDocs: `
+<AspectRatio ratio={16 / 9}>
+  <Image src="..." alt="Image" className="rounded-md object-cover" />
+</AspectRatio>`,
+  },
+  badge: {
+    componentName: 'Badge',
+    importDocs: 'import { Badge } from "/components/ui/badge"',
+    usageDocs: '<Badge>Badge</Badge>',
+  },
+  calendar: {
+    componentName: 'Calendar',
+    importDocs: 'import { Calendar } from "/components/ui/calendar"',
+    usageDocs: '<Calendar />',
+  },
+  carousel: {
+    componentName: 'Carousel',
+    importDocs: `
+import {
+  Carousel,
+  CarouselContent,
+  CarouselItem,
+  CarouselNext,
+  CarouselPrevious,
+} from "/components/ui/carousel"`,
+    usageDocs: `
+<Carousel>
+  <CarouselContent>
+    <CarouselItem>...</CarouselItem>
+    <CarouselItem>...</CarouselItem>
+    <CarouselItem>...</CarouselItem>
+  </CarouselContent>
+  <CarouselPrevious />
+  <CarouselNext />
+</Carousel>`,
+  },
+  collapsible: {
+    componentName: 'Collapsible',
+    importDocs: `
+import {
+  Collapsible,
+  CollapsibleContent,
+  CollapsibleTrigger,
+} from "/components/ui/collapsible"`,
+    usageDocs: `
+<Collapsible>
+  <CollapsibleTrigger>Can I use this in my project?</CollapsibleTrigger>
+  <CollapsibleContent>
+    Yes. Free to use for personal and commercial projects. No attribution required.
+  </CollapsibleContent>
+</Collapsible>`,
+  },
+  dialog: {
+    componentName: 'Dialog',
+    importDocs: `
+import {
+  Dialog,
+  DialogContent,
+  DialogDescription,
+  DialogHeader,
+  DialogTitle,
+  DialogTrigger,
+} from "/components/ui/dialog"`,
+    usageDocs: `
+<Dialog>
+  <DialogTrigger>Open</DialogTrigger>
+  <DialogContent>
+    <DialogHeader>
+      <DialogTitle>Are you sure absolutely sure?</DialogTitle>
+      <DialogDescription>
+        This action cannot be undone.
+      </DialogDescription>
+    </DialogHeader>
+  </DialogContent>
+</Dialog>`,
+  },
+  dropdownMenu: {
+    componentName: 'DropdownMenu',
+    importDocs: `
+import {
+  DropdownMenu,
+  DropdownMenuContent,
+  DropdownMenuItem,
+  DropdownMenuLabel,
+  DropdownMenuSeparator,
+  DropdownMenuTrigger,
+} from "/components/ui/dropdown-menu"`,
+    usageDocs: `
+<DropdownMenu>
+  <DropdownMenuTrigger>Open</DropdownMenuTrigger>
+  <DropdownMenuContent>
+    <DropdownMenuLabel>My Account</DropdownMenuLabel>
+    <DropdownMenuSeparator />
+    <DropdownMenuItem>Profile</DropdownMenuItem>
+    <DropdownMenuItem>Billing</DropdownMenuItem>
+    <DropdownMenuItem>Team</DropdownMenuItem>
+    <DropdownMenuItem>Subscription</DropdownMenuItem>
+  </DropdownMenuContent>
+</DropdownMenu>`,
+  },
+  menubar: {
+    componentName: 'Menubar',
+    importDocs: `
+import {
+  Menubar,
+  MenubarContent,
+  MenubarItem,
+  MenubarMenu,
+  MenubarSeparator,
+  MenubarShortcut,
+  MenubarTrigger,
+} from "/components/ui/menubar"`,
+    usageDocs: `
+<Menubar>
+  <MenubarMenu>
+    <MenubarTrigger>File</MenubarTrigger>
+    <MenubarContent>
+      <MenubarItem>
+        New Tab <MenubarShortcut>⌘T</MenubarShortcut>
+      </MenubarItem>
+      <MenubarItem>New Window</MenubarItem>
+      <MenubarSeparator />
+      <MenubarItem>Share</MenubarItem>
+      <MenubarSeparator />
+      <MenubarItem>Print</MenubarItem>
+    </MenubarContent>
+  </MenubarMenu>
+</Menubar>`,
+  },
+  navigationMenu: {
+    componentName: 'NavigationMenu',
+    importDocs: `
+import {
+  NavigationMenu,
+  NavigationMenuContent,
+  NavigationMenuItem,
+  NavigationMenuLink,
+  NavigationMenuList,
+  NavigationMenuTrigger,
+  navigationMenuTriggerStyle,
+} from "/components/ui/navigation-menu"`,
+    usageDocs: `
+<NavigationMenu>
+  <NavigationMenuList>
+    <NavigationMenuItem>
+      <NavigationMenuTrigger>Item One</NavigationMenuTrigger>
+      <NavigationMenuContent>
+        <NavigationMenuLink>Link</NavigationMenuLink>
+      </NavigationMenuContent>
+    </NavigationMenuItem>
+  </NavigationMenuList>
+</NavigationMenu>`,
+  },
+  popover: {
+    componentName: 'Popover',
+    importDocs: `
+import {
+  Popover,
+  PopoverContent,
+  PopoverTrigger,
+} from "/components/ui/popover"`,
+    usageDocs: `
+<Popover>
+  <PopoverTrigger>Open</PopoverTrigger>
+  <PopoverContent>Place content for the popover here.</PopoverContent>
+</Popover>`,
+  },
+  progress: {
+    componentName: 'Progress',
+    importDocs: 'import { Progress } from "/components/ui/progress"',
+    usageDocs: '<Progress value={33} />',
+  },
+  separator: {
+    componentName: 'Separator',
+    importDocs: 'import { Separator } from "/components/ui/separator"',
+    usageDocs: '<Separator />',
+  },
+  sheet: {
+    componentName: 'Sheet',
+    importDocs: `
+import {
+  Sheet,
+  SheetContent,
+  SheetDescription,
+  SheetHeader,
+  SheetTitle,
+  SheetTrigger,
+} from "/components/ui/sheet"`,
+    usageDocs: `
+<Sheet>
+  <SheetTrigger>Open</SheetTrigger>
+  <SheetContent>
+    <SheetHeader>
+      <SheetTitle>Are you sure absolutely sure?</SheetTitle>
+      <SheetDescription>
+        This action cannot be undone.
+      </SheetDescription>
+    </SheetHeader>
+  </SheetContent>
+</Sheet>`,
+  },
+  skeleton: {
+    componentName: 'Skeleton',
+    importDocs: 'import { Skeleton } from "/components/ui/skeleton"',
+    usageDocs: '<Skeleton className="w-[100px] h-[20px] rounded-full" />',
+  },
+  slider: {
+    componentName: 'Slider',
+    importDocs: 'import { Slider } from "/components/ui/slider"',
+    usageDocs: '<Slider defaultValue={[33]} max={100} step={1} />',
+  },
+  switch: {
+    componentName: 'Switch',
+    importDocs: 'import { Switch } from "/components/ui/switch"',
+    usageDocs: '<Switch />',
+  },
+  table: {
+    componentName: 'Table',
+    importDocs: `
+import {
+  Table,
+  TableBody,
+  TableCaption,
+  TableCell,
+  TableHead,
+  TableHeader,
+  TableRow,
+} from "/components/ui/table"`,
+    usageDocs: `
+<Table>
+  <TableCaption>A list of your recent invoices.</TableCaption>
+  <TableHeader>
+    <TableRow>
+      <TableHead className="w-[100px]">Invoice</TableHead>
+      <TableHead>Status</TableHead>
+      <TableHead>Method</TableHead>
+      <TableHead className="text-right">Amount</TableHead>
+    </TableRow>
+  </TableHeader>
+  <TableBody>
+    <TableRow>
+      <TableCell className="font-medium">INV001</TableCell>
+      <TableCell>Paid</TableCell>
+      <TableCell>Credit Card</TableCell>
+      <TableCell className="text-right">$250.00</TableCell>
+    </TableRow>
+  </TableBody>
+</Table>`,
+  },
+  tabs: {
+    componentName: 'Tabs',
+    importDocs: `
+import {
+  Tabs,
+  TabsContent,
+  TabsList,
+  TabsTrigger,
+} from "/components/ui/tabs"`,
+    usageDocs: `
+<Tabs defaultValue="account" className="w-[400px]">
+  <TabsList>
+    <TabsTrigger value="account">Account</TabsTrigger>
+    <TabsTrigger value="password">Password</TabsTrigger>
+  </TabsList>
+  <TabsContent value="account">Make changes to your account here.</TabsContent>
+  <TabsContent value="password">Change your password here.</TabsContent>
+</Tabs>`,
+  },
+  toast: {
+    componentName: 'Toast',
+    importDocs: `
+import { useToast } from "/components/ui/use-toast"
+import { Button } from "/components/ui/button"`,
+    usageDocs: `
+export function ToastDemo() {
+  const { toast } = useToast()
+  return (
+    <Button
+      onClick={() => {
+        toast({
+          title: "Scheduled: Catch up",
+          description: "Friday, February 10, 2023 at 5:57 PM",
+        })
+      }}
+    >
+      Show Toast
+    </Button>
+  )
+}`,
+  },
+  toggle: {
+    componentName: 'Toggle',
+    importDocs: 'import { Toggle } from "/components/ui/toggle"',
+    usageDocs: '<Toggle>Toggle</Toggle>',
+  },
+  tooltip: {
+    componentName: 'Tooltip',
+    importDocs: `
+import {
+  Tooltip,
+  TooltipContent,
+  TooltipProvider,
+  TooltipTrigger,
+} from "/components/ui/tooltip"`,
+    usageDocs: `
+<TooltipProvider>
+  <Tooltip>
+    <TooltipTrigger>Hover</TooltipTrigger>
+    <TooltipContent>
+      <p>Add to library</p>
+    </TooltipContent>
+  </Tooltip>
+</TooltipProvider>`,
+  },
+};
+
+const components = Object.assign({}, essentialComponents, extraComponents);
+
+module.exports = {
+  components,
+};
--- a/api/app/clients/prompts/shadcn-docs/generate.js
+++ b/api/app/clients/prompts/shadcn-docs/generate.js
@@ -0,0 +1,50 @@
+const dedent = require('dedent');
+
+/**
+ * Generate system prompt for AI-assisted React component creation
+ * @param {Object} options - Configuration options
+ * @param {Object} options.components - Documentation for shadcn components
+ * @param {boolean} [options.useXML=false] - Whether to use XML-style formatting for component instructions
+ * @returns {string} The generated system prompt
+ */
+function generateShadcnPrompt(options) {
+  const { components, useXML = false } = options;
+
+  let systemPrompt = dedent`
+    ## Additional Artifact Instructions for React Components: "application/vnd.react"
+    
+    There are some prestyled components (primitives) available for use. Please use your best judgement to use any of these components if the app calls for one.
+
+    Here are the components that are available, along with how to import them, and how to use them:
+
+    ${Object.values(components)
+    .map((component) => {
+      if (useXML) {
+        return dedent`
+            <component>
+              <name>${component.componentName}</name>
+              <import-instructions>${component.importDocs}</import-instructions>
+              <usage-instructions>${component.usageDocs}</usage-instructions>
+            </component>
+          `;
+      } else {
+        return dedent`
+            # ${component.componentName}
+
+            ## Import Instructions
+            ${component.importDocs}
+
+            ## Usage Instructions
+            ${component.usageDocs}
+          `;
+      }
+    })
+    .join('\n\n')}
+  `;
+
+  return systemPrompt;
+}
+
+module.exports = {
+  generateShadcnPrompt,
+};
--- a/api/app/clients/prompts/summaryPrompts.js
+++ b/api/app/clients/prompts/summaryPrompts.js
@@ -0,0 +1,53 @@
+const { PromptTemplate } = require('langchain/prompts');
+/*
+ * Without `{summary}` and `{new_lines}`, token count is 98
+ * We are counting this towards the max context tokens for summaries, +3 for the assistant label (101)
+ * If this prompt changes, use https://tiktokenizer.vercel.app/ to count the tokens
+ */
+const _DEFAULT_SUMMARIZER_TEMPLATE = `Summarize the conversation by integrating new lines into the current summary.
+
+EXAMPLE:
+Current summary:
+The human inquires about the AI's view on artificial intelligence. The AI believes it's beneficial.
+
+New lines:
+Human: Why is it beneficial?
+AI: It helps humans achieve their potential.
+
+New summary:
+The human inquires about the AI's view on artificial intelligence. The AI believes it's beneficial because it helps humans achieve their potential.
+
+Current summary:
+{summary}
+
+New lines:
+{new_lines}
+
+New summary:`;
+
+const SUMMARY_PROMPT = new PromptTemplate({
+  inputVariables: ['summary', 'new_lines'],
+  template: _DEFAULT_SUMMARIZER_TEMPLATE,
+});
+
+/*
+ * Without `{new_lines}`, token count is 27
+ * We are counting this towards the max context tokens for summaries, rounded up to 30
+ * If this prompt changes, use https://tiktokenizer.vercel.app/ to count the tokens
+ */
+const _CUT_OFF_SUMMARIZER = `The following text is cut-off:
+{new_lines}
+
+Summarize the content as best as you can, noting that it was cut-off.
+
+Summary:`;
+
+const CUT_OFF_PROMPT = new PromptTemplate({
+  inputVariables: ['new_lines'],
+  template: _CUT_OFF_SUMMARIZER,
+});
+
+module.exports = {
+  SUMMARY_PROMPT,
+  CUT_OFF_PROMPT,
+};
--- a/api/app/clients/prompts/titlePrompts.js
+++ b/api/app/clients/prompts/titlePrompts.js
@@ -16,7 +16,7 @@ const createTitlePrompt = ({ convo }) => {
  const titlePrompt = new ChatPromptTemplate({
    promptMessages: [
      SystemMessagePromptTemplate.fromTemplate(
-        `Write a concise title for this conversation in the given language. Title in 5 Words or Less. No Punctuation or Quotation. All first letters of every word must be capitalized (resembling title-case), written in the given Language.
+        `Write a concise title for this conversation in the given language. Title in 5 Words or Less. No Punctuation or Quotation. Must be in Title Case, written in the given Language.
 ${convo}`,
      ),
      HumanMessagePromptTemplate.fromTemplate('Language: {language}'),
@@ -27,7 +27,96 @@ ${convo}`,
  return titlePrompt;
 };

+const titleInstruction =
+  'a concise, 5-word-or-less title for the conversation, using its same language, with no punctuation. Apply title case conventions appropriate for the language. Never directly mention the language name or the word "title"';
+const titleFunctionPrompt = `In this environment you have access to a set of tools you can use to generate the conversation title.
+  
+You may call them like this:
+<function_calls>
+<invoke>
+<tool_name>$TOOL_NAME</tool_name>
+<parameters>
+<$PARAMETER_NAME>$PARAMETER_VALUE</$PARAMETER_NAME>
+...
+</parameters>
+</invoke>
+</function_calls>
+
+Here are the tools available:
+<tools>
+<tool_description>
+<tool_name>submit_title</tool_name>
+<description>
+Submit a brief title in the conversation's language, following the parameter description closely.
+</description>
+<parameters>
+<parameter>
+<name>title</name>
+<type>string</type>
+<description>${titleInstruction}</description>
+</parameter>
+</parameters>
+</tool_description>
+</tools>`;
+
+const genTranslationPrompt = (
+  translationPrompt,
+) => `In this environment you have access to a set of tools you can use to translate text.
+  
+You may call them like this:
+<function_calls>
+<invoke>
+<tool_name>$TOOL_NAME</tool_name>
+<parameters>
+<$PARAMETER_NAME>$PARAMETER_VALUE</$PARAMETER_NAME>
+...
+</parameters>
+</invoke>
+</function_calls>
+
+Here are the tools available:
+<tools>
+<tool_description>
+<tool_name>submit_translation</tool_name>
+<description>
+Submit a translation in the target language, following the parameter description and its language closely.
+</description>
+<parameters>
+<parameter>
+<name>translation</name>
+<type>string</type>
+<description>${translationPrompt}
+ONLY include the generated translation without quotations, nor its related key</description>
+</parameter>
+</parameters>
+</tool_description>
+</tools>`;
+
+/**
+ * Parses specified parameter from the provided prompt.
+ * @param {string} prompt - The prompt containing the desired parameter.
+ * @param {string} paramName - The name of the parameter to extract.
+ * @returns {string} The parsed parameter's value or a default value if not found.
+ */
+function parseParamFromPrompt(prompt, paramName) {
+  const paramRegex = new RegExp(`<${paramName}>([\\s\\S]+?)</${paramName}>`);
+  const paramMatch = prompt.match(paramRegex);
+
+  if (paramMatch && paramMatch[1]) {
+    return paramMatch[1].trim();
+  }
+
+  if (prompt && prompt.length) {
+    return `NO TOOL INVOCATION: ${prompt}`;
+  }
+  return `No ${paramName} provided`;
+}
+
 module.exports = {
  langPrompt,
+  titleInstruction,
  createTitlePrompt,
+  titleFunctionPrompt,
+  parseParamFromPrompt,
+  genTranslationPrompt,
 };
--- a/api/app/clients/prompts/truncateText.js
+++ b/api/app/clients/prompts/truncateText.js
@@ -0,0 +1,40 @@
+const MAX_CHAR = 255;
+
+/**
+ * Truncates a given text to a specified maximum length, appending ellipsis and a notification
+ * if the original text exceeds the maximum length.
+ *
+ * @param {string} text - The text to be truncated.
+ * @param {number} [maxLength=MAX_CHAR] - The maximum length of the text after truncation. Defaults to MAX_CHAR.
+ * @returns {string} The truncated text if the original text length exceeds maxLength, otherwise returns the original text.
+ */
+function truncateText(text, maxLength = MAX_CHAR) {
+  if (text.length > maxLength) {
+    return `${text.slice(0, maxLength)}... [text truncated for brevity]`;
+  }
+  return text;
+}
+
+/**
+ * Truncates a given text to a specified maximum length by showing the first half and the last half of the text,
+ * separated by ellipsis. This method ensures the output does not exceed the maximum length, including the addition
+ * of ellipsis and notification if the original text exceeds the maximum length.
+ *
+ * @param {string} text - The text to be truncated.
+ * @param {number} [maxLength=MAX_CHAR] - The maximum length of the output text after truncation. Defaults to MAX_CHAR.
+ * @returns {string} The truncated text showing the first half and the last half, or the original text if it does not exceed maxLength.
+ */
+function smartTruncateText(text, maxLength = MAX_CHAR) {
+  const ellipsis = '...';
+  const notification = ' [text truncated for brevity]';
+  const halfMaxLength = Math.floor((maxLength - ellipsis.length - notification.length) / 2);
+
+  if (text.length > maxLength) {
+    const startLastHalf = text.length - halfMaxLength;
+    return `${text.slice(0, halfMaxLength)}${ellipsis}${text.slice(startLastHalf)}${notification}`;
+  }
+
+  return text;
+}
+
+module.exports = { truncateText, smartTruncateText };
--- a/api/app/clients/specs/AnthropicClient.test.js
+++ b/api/app/clients/specs/AnthropicClient.test.js
@@ -1,4 +1,6 @@
-const AnthropicClient = require('../AnthropicClient');
+const { anthropicSettings } = require('librechat-data-provider');
+const AnthropicClient = require('~/app/clients/AnthropicClient');
+
 const HUMAN_PROMPT = '\n\nHuman:';
 const AI_PROMPT = '\n\nAssistant:';

@@ -22,7 +24,7 @@ describe('AnthropicClient', () => {
    const options = {
      modelOptions: {
        model,
-        temperature: 0.7,
+        temperature: anthropicSettings.temperature.default,
      },
    };
    client = new AnthropicClient('test-api-key');
@@ -33,7 +35,42 @@ describe('AnthropicClient', () => {
    it('should set the options correctly', () => {
      expect(client.apiKey).toBe('test-api-key');
      expect(client.modelOptions.model).toBe(model);
-      expect(client.modelOptions.temperature).toBe(0.7);
+      expect(client.modelOptions.temperature).toBe(anthropicSettings.temperature.default);
+    });
+
+    it('should set legacy maxOutputTokens for non-Claude-3 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-2',
+          maxOutputTokens: anthropicSettings.maxOutputTokens.default,
+        },
+      });
+      expect(client.modelOptions.maxOutputTokens).toBe(
+        anthropicSettings.legacy.maxOutputTokens.default,
+      );
+    });
+    it('should not set maxOutputTokens if not provided', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3',
+        },
+      });
+      expect(client.modelOptions.maxOutputTokens).toBeUndefined();
+    });
+
+    it('should not set legacy maxOutputTokens for Claude-3 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3-opus-20240229',
+          maxOutputTokens: anthropicSettings.legacy.maxOutputTokens.default,
+        },
+      });
+      expect(client.modelOptions.maxOutputTokens).toBe(
+        anthropicSettings.legacy.maxOutputTokens.default,
+      );
    });
  });

@@ -136,4 +173,212 @@ describe('AnthropicClient', () => {
      expect(prompt).toContain('You are Claude-2');
    });
  });
+
+  describe('getClient', () => {
+    it('should set legacy maxOutputTokens for non-Claude-3 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-2',
+          maxOutputTokens: anthropicSettings.legacy.maxOutputTokens.default,
+        },
+      });
+      expect(client.modelOptions.maxOutputTokens).toBe(
+        anthropicSettings.legacy.maxOutputTokens.default,
+      );
+    });
+
+    it('should not set legacy maxOutputTokens for Claude-3 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3-opus-20240229',
+          maxOutputTokens: anthropicSettings.legacy.maxOutputTokens.default,
+        },
+      });
+      expect(client.modelOptions.maxOutputTokens).toBe(
+        anthropicSettings.legacy.maxOutputTokens.default,
+      );
+    });
+
+    it('should add beta header for claude-3-5-sonnet model', () => {
+      const client = new AnthropicClient('test-api-key');
+      const modelOptions = {
+        model: 'claude-3-5-sonnet-20240307',
+      };
+      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(
+        'max-tokens-3-5-sonnet-2024-07-15,prompt-caching-2024-07-31',
+      );
+    });
+
+    it('should add beta header for claude-3-haiku model', () => {
+      const client = new AnthropicClient('test-api-key');
+      const modelOptions = {
+        model: 'claude-3-haiku-2028',
+      };
+      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 other models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-2',
+        },
+      });
+      const anthropicClient = client.getClient();
+      expect(anthropicClient.defaultHeaders).not.toHaveProperty('anthropic-beta');
+    });
+  });
+
+  describe('calculateCurrentTokenCount', () => {
+    let client;
+
+    beforeEach(() => {
+      client = new AnthropicClient('test-api-key');
+    });
+
+    it('should calculate correct token count when usage is provided', () => {
+      const tokenCountMap = {
+        msg1: 10,
+        msg2: 20,
+        currentMsg: 30,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        input_tokens: 70,
+        output_tokens: 50,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(40); // 70 - (10 + 20) = 40
+    });
+
+    it('should return original estimate if calculation results in negative value', () => {
+      const tokenCountMap = {
+        msg1: 40,
+        msg2: 50,
+        currentMsg: 30,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        input_tokens: 80,
+        output_tokens: 50,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(30); // Original estimate
+    });
+
+    it('should handle cache creation and read input tokens', () => {
+      const tokenCountMap = {
+        msg1: 10,
+        msg2: 20,
+        currentMsg: 30,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        input_tokens: 50,
+        cache_creation_input_tokens: 10,
+        cache_read_input_tokens: 20,
+        output_tokens: 40,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(50); // (50 + 10 + 20) - (10 + 20) = 50
+    });
+
+    it('should handle missing usage properties', () => {
+      const tokenCountMap = {
+        msg1: 10,
+        msg2: 20,
+        currentMsg: 30,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        output_tokens: 40,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(30); // Original estimate
+    });
+
+    it('should handle empty tokenCountMap', () => {
+      const tokenCountMap = {};
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        input_tokens: 50,
+        output_tokens: 40,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(50);
+      expect(Number.isNaN(result)).toBe(false);
+    });
+
+    it('should handle zero values in usage', () => {
+      const tokenCountMap = {
+        msg1: 10,
+        currentMsg: 20,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        input_tokens: 0,
+        cache_creation_input_tokens: 0,
+        cache_read_input_tokens: 0,
+        output_tokens: 0,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(20); // Should return original estimate
+      expect(Number.isNaN(result)).toBe(false);
+    });
+
+    it('should handle undefined usage', () => {
+      const tokenCountMap = {
+        msg1: 10,
+        currentMsg: 20,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = undefined;
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(20); // Should return original estimate
+      expect(Number.isNaN(result)).toBe(false);
+    });
+
+    it('should handle non-numeric values in tokenCountMap', () => {
+      const tokenCountMap = {
+        msg1: 'ten',
+        currentMsg: 20,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        input_tokens: 30,
+        output_tokens: 10,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(30); // Should return 30 (input_tokens) - 0 (ignored 'ten') = 30
+      expect(Number.isNaN(result)).toBe(false);
+    });
+  });
 });
--- a/api/app/clients/specs/BaseClient.test.js
+++ b/api/app/clients/specs/BaseClient.test.js
@@ -1,27 +1,34 @@
+const { Constants } = require('librechat-data-provider');
 const { initializeFakeClient } = require('./FakeClient');

-jest.mock('../../../lib/db/connectDb');
-jest.mock('../../../models', () => {
-  return function () {
-    return {
-      save: jest.fn(),
-      deleteConvos: jest.fn(),
-      getConvo: jest.fn(),
-      getMessages: jest.fn(),
-      saveMessage: jest.fn(),
-      updateMessage: jest.fn(),
-      saveConvo: jest.fn(),
-    };
-  };
-});
-
-jest.mock('langchain/text_splitter', () => {
-  return {
-    RecursiveCharacterTextSplitter: jest.fn().mockImplementation(() => {
-      return { createDocuments: jest.fn().mockResolvedValue([]) };
-    }),
-  };
-});
+jest.mock('~/lib/db/connectDb');
+jest.mock('~/models', () => ({
+  User: jest.fn(),
+  Key: jest.fn(),
+  Session: jest.fn(),
+  Balance: jest.fn(),
+  Transaction: jest.fn(),
+  getMessages: jest.fn().mockResolvedValue([]),
+  saveMessage: jest.fn(),
+  updateMessage: jest.fn(),
+  deleteMessagesSince: jest.fn(),
+  deleteMessages: jest.fn(),
+  getConvoTitle: jest.fn(),
+  getConvo: jest.fn(),
+  saveConvo: jest.fn(),
+  deleteConvos: jest.fn(),
+  getPreset: jest.fn(),
+  getPresets: jest.fn(),
+  savePreset: jest.fn(),
+  deletePresets: jest.fn(),
+  findFileById: jest.fn(),
+  createFile: jest.fn(),
+  updateFile: jest.fn(),
+  deleteFile: jest.fn(),
+  deleteFiles: jest.fn(),
+  getFiles: jest.fn(),
+  updateFileUsage: jest.fn(),
+}));

 jest.mock('langchain/chat_models/openai', () => {
  return {
@@ -31,14 +38,6 @@ jest.mock('langchain/chat_models/openai', () => {
  };
 });

-jest.mock('langchain/chains', () => {
-  return {
-    loadSummarizationChain: jest.fn().mockReturnValue({
-      call: jest.fn().mockResolvedValue({ output_text: 'Refined answer' }),
-    }),
-  };
-});
-
 let parentMessageId;
 let conversationId;
 const fakeMessages = [];
@@ -69,6 +68,13 @@ describe('BaseClient', () => {

  beforeEach(() => {
    TestClient = initializeFakeClient(apiKey, options, fakeMessages);
+    TestClient.summarizeMessages = jest.fn().mockResolvedValue({
+      summaryMessage: {
+        role: 'system',
+        content: 'Refined answer',
+      },
+      summaryTokenCount: 5,
+    });
  });

  test('returns the input messages without instructions when addInstructions() is called with empty instructions', () => {
@@ -103,30 +109,24 @@ describe('BaseClient', () => {
    expect(result).toBe(expected);
  });

-  test('refines messages correctly in refineMessages()', async () => {
+  test('refines messages correctly in summarizeMessages()', async () => {
    const messagesToRefine = [
      { role: 'user', content: 'Hello', tokenCount: 10 },
      { role: 'assistant', content: 'How can I help you?', tokenCount: 20 },
    ];
    const remainingContextTokens = 100;
    const expectedRefinedMessage = {
-      role: 'assistant',
+      role: 'system',
      content: 'Refined answer',
-      tokenCount: 14, // 'Refined answer'.length
    };

-    const result = await TestClient.refineMessages(messagesToRefine, remainingContextTokens);
-    expect(result).toEqual(expectedRefinedMessage);
+    const result = await TestClient.summarizeMessages({ messagesToRefine, remainingContextTokens });
+    expect(result.summaryMessage).toEqual(expectedRefinedMessage);
  });

  test('gets messages within token limit (under limit) correctly in getMessagesWithinTokenLimit()', async () => {
    TestClient.maxContextTokens = 100;
-    TestClient.shouldRefineContext = true;
-    TestClient.refineMessages = jest.fn().mockResolvedValue({
-      role: 'assistant',
-      content: 'Refined answer',
-      tokenCount: 30,
-    });
+    TestClient.shouldSummarize = true;

    const messages = [
      { role: 'user', content: 'Hello', tokenCount: 5 },
@@ -142,43 +142,50 @@ describe('BaseClient', () => {
    const expectedRemainingContextTokens = 58 - 3; // (100 - 5 - 19 - 18) - 3
    const expectedMessagesToRefine = [];

+    const lastExpectedMessage =
+      expectedMessagesToRefine?.[expectedMessagesToRefine.length - 1] ?? {};
+    const expectedIndex = messages.findIndex((msg) => msg.content === lastExpectedMessage?.content);
+
    const result = await TestClient.getMessagesWithinTokenLimit(messages);
+
    expect(result.context).toEqual(expectedContext);
+    expect(result.summaryIndex).toEqual(expectedIndex);
    expect(result.remainingContextTokens).toBe(expectedRemainingContextTokens);
    expect(result.messagesToRefine).toEqual(expectedMessagesToRefine);
  });

-  test('gets messages within token limit (over limit) correctly in getMessagesWithinTokenLimit()', async () => {
+  test('gets result over token limit correctly in getMessagesWithinTokenLimit()', async () => {
    TestClient.maxContextTokens = 50; // Set a lower limit
-    TestClient.shouldRefineContext = true;
-    TestClient.refineMessages = jest.fn().mockResolvedValue({
-      role: 'assistant',
-      content: 'Refined answer',
-      tokenCount: 4,
-    });
+    TestClient.shouldSummarize = true;

    const messages = [
-      { role: 'user', content: 'I need a coffee, stat!', tokenCount: 30 },
-      { role: 'assistant', content: 'Sure, I can help with that.', tokenCount: 30 },
-      { role: 'user', content: 'Hello', tokenCount: 5 },
-      { role: 'assistant', content: 'How can I help you?', tokenCount: 19 },
-      { role: 'user', content: 'I have a question.', tokenCount: 18 },
-    ];
-    const expectedContext = [
-      { role: 'user', content: 'Hello', tokenCount: 5 },
-      { role: 'assistant', content: 'How can I help you?', tokenCount: 19 },
-      { role: 'user', content: 'I have a question.', tokenCount: 18 },
+      { role: 'user', content: 'Hello', tokenCount: 30 },
+      { role: 'assistant', content: 'How can I help you?', tokenCount: 30 },
+      { role: 'user', content: 'I have a question.', tokenCount: 5 },
+      { role: 'user', content: 'I need a coffee, stat!', tokenCount: 19 },
+      { role: 'assistant', content: 'Sure, I can help with that.', tokenCount: 18 },
    ];

    // Subtract 3 tokens for Assistant Label priming after all messages have been counted.
-    const expectedRemainingContextTokens = 8 - 3; // (50 - 18 - 19 - 5) - 3
+    const expectedRemainingContextTokens = 5; // (50 - 18 - 19 - 5) - 3
    const expectedMessagesToRefine = [
-      { role: 'user', content: 'I need a coffee, stat!', tokenCount: 30 },
-      { role: 'assistant', content: 'Sure, I can help with that.', tokenCount: 30 },
+      { role: 'user', content: 'Hello', tokenCount: 30 },
+      { role: 'assistant', content: 'How can I help you?', tokenCount: 30 },
+    ];
+    const expectedContext = [
+      { role: 'user', content: 'I have a question.', tokenCount: 5 },
+      { role: 'user', content: 'I need a coffee, stat!', tokenCount: 19 },
+      { role: 'assistant', content: 'Sure, I can help with that.', tokenCount: 18 },
    ];

+    const lastExpectedMessage =
+      expectedMessagesToRefine?.[expectedMessagesToRefine.length - 1] ?? {};
+    const expectedIndex = messages.findIndex((msg) => msg.content === lastExpectedMessage?.content);
+
    const result = await TestClient.getMessagesWithinTokenLimit(messages);
+
    expect(result.context).toEqual(expectedContext);
+    expect(result.summaryIndex).toEqual(expectedIndex);
    expect(result.remainingContextTokens).toBe(expectedRemainingContextTokens);
    expect(result.messagesToRefine).toEqual(expectedMessagesToRefine);
  });
@@ -200,14 +207,10 @@ describe('BaseClient', () => {
      ],
      remainingContextTokens: 80,
      messagesToRefine: [{ content: 'Hello' }],
-      refineIndex: 3,
+      summaryIndex: 3,
    });
-    TestClient.refineMessages = jest.fn().mockResolvedValue({
-      role: 'assistant',
-      content: 'Refined answer',
-      tokenCount: 30,
-    });
-    TestClient.getTokenCountForResponse = jest.fn().mockReturnValue(40);
+
+    TestClient.getTokenCount = jest.fn().mockReturnValue(40);

    const instructions = { content: 'Please provide more details.' };
    const orderedMessages = [
@@ -225,9 +228,8 @@ describe('BaseClient', () => {
    const expectedResult = {
      payload: [
        {
+          role: 'system',
          content: 'Refined answer',
-          role: 'assistant',
-          tokenCount: 30,
        },
        { content: 'How can I help you?' },
        { content: 'Please provide more details.' },
@@ -238,14 +240,214 @@ describe('BaseClient', () => {
      messages: expect.any(Array),
    };

+    TestClient.shouldSummarize = true;
    const result = await TestClient.handleContextStrategy({
      instructions,
      orderedMessages,
      formattedMessages,
    });
+
    expect(result).toEqual(expectedResult);
  });

+  describe('getMessagesForConversation', () => {
+    it('should return an empty array if the parentMessageId does not exist', () => {
+      const result = TestClient.constructor.getMessagesForConversation({
+        messages: unorderedMessages,
+        parentMessageId: '999',
+      });
+      expect(result).toEqual([]);
+    });
+
+    it('should handle messages with messageId property', () => {
+      const messagesWithMessageId = [
+        { messageId: '1', parentMessageId: null, text: 'Message 1' },
+        { messageId: '2', parentMessageId: '1', text: 'Message 2' },
+      ];
+      const result = TestClient.constructor.getMessagesForConversation({
+        messages: messagesWithMessageId,
+        parentMessageId: '2',
+      });
+      expect(result).toEqual([
+        { messageId: '1', parentMessageId: null, text: 'Message 1' },
+        { messageId: '2', parentMessageId: '1', text: 'Message 2' },
+      ]);
+    });
+
+    const messagesWithNullParent = [
+      { id: '1', parentMessageId: null, text: 'Message 1' },
+      { id: '2', parentMessageId: null, text: 'Message 2' },
+    ];
+
+    it('should handle messages with null parentMessageId that are not root', () => {
+      const result = TestClient.constructor.getMessagesForConversation({
+        messages: messagesWithNullParent,
+        parentMessageId: '2',
+      });
+      expect(result).toEqual([{ id: '2', parentMessageId: null, text: 'Message 2' }]);
+    });
+
+    const cyclicMessages = [
+      { id: '3', parentMessageId: '2', text: 'Message 3' },
+      { id: '1', parentMessageId: '3', text: 'Message 1' },
+      { id: '2', parentMessageId: '1', text: 'Message 2' },
+    ];
+
+    it('should handle cyclic references without going into an infinite loop', () => {
+      const result = TestClient.constructor.getMessagesForConversation({
+        messages: cyclicMessages,
+        parentMessageId: '3',
+      });
+      expect(result).toEqual([
+        { id: '1', parentMessageId: '3', text: 'Message 1' },
+        { id: '2', parentMessageId: '1', text: 'Message 2' },
+        { id: '3', parentMessageId: '2', text: 'Message 3' },
+      ]);
+    });
+
+    const unorderedMessages = [
+      { id: '3', parentMessageId: '2', text: 'Message 3' },
+      { id: '2', parentMessageId: '1', text: 'Message 2' },
+      { id: '1', parentMessageId: Constants.NO_PARENT, text: 'Message 1' },
+    ];
+
+    it('should return ordered messages based on parentMessageId', () => {
+      const result = TestClient.constructor.getMessagesForConversation({
+        messages: unorderedMessages,
+        parentMessageId: '3',
+      });
+      expect(result).toEqual([
+        { id: '1', parentMessageId: Constants.NO_PARENT, text: 'Message 1' },
+        { id: '2', parentMessageId: '1', text: 'Message 2' },
+        { id: '3', parentMessageId: '2', text: 'Message 3' },
+      ]);
+    });
+
+    const unorderedBranchedMessages = [
+      { id: '4', parentMessageId: '2', text: 'Message 4', summary: 'Summary for Message 4' },
+      { id: '10', parentMessageId: '7', text: 'Message 10' },
+      { id: '1', parentMessageId: null, text: 'Message 1' },
+      { id: '6', parentMessageId: '5', text: 'Message 7' },
+      { id: '7', parentMessageId: '5', text: 'Message 7' },
+      { id: '2', parentMessageId: '1', text: 'Message 2' },
+      { id: '8', parentMessageId: '6', text: 'Message 8' },
+      { id: '5', parentMessageId: '3', text: 'Message 5' },
+      { id: '3', parentMessageId: '1', text: 'Message 3' },
+      { id: '6', parentMessageId: '4', text: 'Message 6' },
+      { id: '8', parentMessageId: '7', text: 'Message 9' },
+      { id: '9', parentMessageId: '7', text: 'Message 9' },
+      { id: '11', parentMessageId: '2', text: 'Message 11', summary: 'Summary for Message 11' },
+    ];
+
+    it('should return ordered messages from a branched array based on parentMessageId', () => {
+      const result = TestClient.constructor.getMessagesForConversation({
+        messages: unorderedBranchedMessages,
+        parentMessageId: '10',
+        summary: true,
+      });
+      expect(result).toEqual([
+        { id: '1', parentMessageId: null, text: 'Message 1' },
+        { id: '3', parentMessageId: '1', text: 'Message 3' },
+        { id: '5', parentMessageId: '3', text: 'Message 5' },
+        { id: '7', parentMessageId: '5', text: 'Message 7' },
+        { id: '10', parentMessageId: '7', text: 'Message 10' },
+      ]);
+    });
+
+    it('should return an empty array if no messages are provided', () => {
+      const result = TestClient.constructor.getMessagesForConversation({
+        messages: [],
+        parentMessageId: '3',
+      });
+      expect(result).toEqual([]);
+    });
+
+    it('should map over the ordered messages if mapMethod is provided', () => {
+      const mapMethod = (msg) => msg.text;
+      const result = TestClient.constructor.getMessagesForConversation({
+        messages: unorderedMessages,
+        parentMessageId: '3',
+        mapMethod,
+      });
+      expect(result).toEqual(['Message 1', 'Message 2', 'Message 3']);
+    });
+
+    let unorderedMessagesWithSummary = [
+      { id: '4', parentMessageId: '3', text: 'Message 4' },
+      { id: '2', parentMessageId: '1', text: 'Message 2', summary: 'Summary for Message 2' },
+      { id: '3', parentMessageId: '2', text: 'Message 3', summary: 'Summary for Message 3' },
+      { id: '1', parentMessageId: null, text: 'Message 1' },
+    ];
+
+    it('should start with the message that has a summary property and continue until the specified parentMessageId', () => {
+      const result = TestClient.constructor.getMessagesForConversation({
+        messages: unorderedMessagesWithSummary,
+        parentMessageId: '4',
+        summary: true,
+      });
+      expect(result).toEqual([
+        {
+          id: '3',
+          parentMessageId: '2',
+          role: 'system',
+          text: 'Summary for Message 3',
+          summary: 'Summary for Message 3',
+        },
+        { id: '4', parentMessageId: '3', text: 'Message 4' },
+      ]);
+    });
+
+    it('should handle multiple summaries and return the branch from the latest to the parentMessageId', () => {
+      unorderedMessagesWithSummary = [
+        { id: '5', parentMessageId: '4', text: 'Message 5' },
+        { id: '2', parentMessageId: '1', text: 'Message 2', summary: 'Summary for Message 2' },
+        { id: '3', parentMessageId: '2', text: 'Message 3', summary: 'Summary for Message 3' },
+        { id: '4', parentMessageId: '3', text: 'Message 4', summary: 'Summary for Message 4' },
+        { id: '1', parentMessageId: null, text: 'Message 1' },
+      ];
+      const result = TestClient.constructor.getMessagesForConversation({
+        messages: unorderedMessagesWithSummary,
+        parentMessageId: '5',
+        summary: true,
+      });
+      expect(result).toEqual([
+        {
+          id: '4',
+          parentMessageId: '3',
+          role: 'system',
+          text: 'Summary for Message 4',
+          summary: 'Summary for Message 4',
+        },
+        { id: '5', parentMessageId: '4', text: 'Message 5' },
+      ]);
+    });
+
+    it('should handle summary at root edge case and continue until the parentMessageId', () => {
+      unorderedMessagesWithSummary = [
+        { id: '5', parentMessageId: '4', text: 'Message 5' },
+        { id: '1', parentMessageId: null, text: 'Message 1', summary: 'Summary for Message 1' },
+        { id: '4', parentMessageId: '3', text: 'Message 4', summary: 'Summary for Message 4' },
+        { id: '2', parentMessageId: '1', text: 'Message 2', summary: 'Summary for Message 2' },
+        { id: '3', parentMessageId: '2', text: 'Message 3', summary: 'Summary for Message 3' },
+      ];
+      const result = TestClient.constructor.getMessagesForConversation({
+        messages: unorderedMessagesWithSummary,
+        parentMessageId: '5',
+        summary: true,
+      });
+      expect(result).toEqual([
+        {
+          id: '4',
+          parentMessageId: '3',
+          role: 'system',
+          text: 'Summary for Message 4',
+          summary: 'Summary for Message 4',
+        },
+        { id: '5', parentMessageId: '4', text: 'Message 5' },
+      ]);
+    });
+  });
+
  describe('sendMessage', () => {
    test('sendMessage should return a response message', async () => {
      const expectedResult = expect.objectContaining({
@@ -268,7 +470,7 @@ describe('BaseClient', () => {
      const opts = {
        conversationId,
        parentMessageId,
-        getIds: jest.fn(),
+        getReqData: jest.fn(),
        onStart: jest.fn(),
      };

@@ -285,7 +487,7 @@ describe('BaseClient', () => {
      parentMessageId = response.messageId;
      expect(response.conversationId).toEqual(conversationId);
      expect(response).toEqual(expectedResult);
-      expect(opts.getIds).toHaveBeenCalled();
+      expect(opts.getReqData).toHaveBeenCalled();
      expect(opts.onStart).toHaveBeenCalled();
      expect(TestClient.getBuildMessagesOptions).toHaveBeenCalled();
      expect(TestClient.getSaveOptions).toHaveBeenCalled();
@@ -342,9 +544,9 @@ describe('BaseClient', () => {
      );
    });

-    test('setOptions is called with the correct arguments', async () => {
+    test('setOptions is called with the correct arguments only when replaceOptions is set to true', async () => {
      TestClient.setOptions = jest.fn();
-      const opts = { conversationId: '123', parentMessageId: '456' };
+      const opts = { conversationId: '123', parentMessageId: '456', replaceOptions: true };
      await TestClient.sendMessage('Hello, world!', opts);
      expect(TestClient.setOptions).toHaveBeenCalledWith(opts);
      TestClient.setOptions.mockClear();
@@ -359,11 +561,11 @@ describe('BaseClient', () => {
      );
    });

-    test('getIds is called with the correct arguments', async () => {
-      const getIds = jest.fn();
-      const opts = { getIds };
+    test('getReqData is called with the correct arguments', async () => {
+      const getReqData = jest.fn();
+      const opts = { getReqData };
      const response = await TestClient.sendMessage('Hello, world!', opts);
-      expect(getIds).toHaveBeenCalledWith({
+      expect(getReqData).toHaveBeenCalledWith({
        userMessage: expect.objectContaining({ text: 'Hello, world!' }),
        conversationId: response.conversationId,
        responseMessageId: response.messageId,
@@ -374,7 +576,11 @@ describe('BaseClient', () => {
      const onStart = jest.fn();
      const opts = { onStart };
      await TestClient.sendMessage('Hello, world!', opts);
-      expect(onStart).toHaveBeenCalledWith(expect.objectContaining({ text: 'Hello, world!' }));
+
+      expect(onStart).toHaveBeenCalledWith(
+        expect.objectContaining({ text: 'Hello, world!' }),
+        expect.any(String),
+      );
    });

    test('saveMessageToDatabase is called with the correct arguments', async () => {
@@ -404,12 +610,12 @@ describe('BaseClient', () => {
      expect(TestClient.sendCompletion).toHaveBeenCalledWith(payload, opts);
    });

-    test('getTokenCountForResponse is called with the correct arguments', async () => {
+    test('getTokenCount for response is called with the correct arguments', async () => {
      const tokenCountMap = {}; // Mock tokenCountMap
      TestClient.buildMessages.mockReturnValue({ prompt: [], tokenCountMap });
-      TestClient.getTokenCountForResponse = jest.fn();
+      TestClient.getTokenCount = jest.fn();
      const response = await TestClient.sendMessage('Hello, world!', {});
-      expect(TestClient.getTokenCountForResponse).toHaveBeenCalledWith(response);
+      expect(TestClient.getTokenCount).toHaveBeenCalledWith(response.text);
    });

    test('returns an object with the correct shape', async () => {
@@ -425,5 +631,32 @@ describe('BaseClient', () => {
        }),
      );
    });
+
+    test('userMessagePromise is awaited before saving response message', async () => {
+      // Mock the saveMessageToDatabase method
+      TestClient.saveMessageToDatabase = jest.fn().mockImplementation(() => {
+        return new Promise((resolve) => setTimeout(resolve, 100)); // Simulate a delay
+      });
+
+      // Send a message
+      const messagePromise = TestClient.sendMessage('Hello, world!');
+
+      // Wait a short time to ensure the user message save has started
+      await new Promise((resolve) => setTimeout(resolve, 50));
+
+      // Check that saveMessageToDatabase has been called once (for the user message)
+      expect(TestClient.saveMessageToDatabase).toHaveBeenCalledTimes(1);
+
+      // Wait for the message to be fully processed
+      await messagePromise;
+
+      // Check that saveMessageToDatabase has been called twice (once for user message, once for response)
+      expect(TestClient.saveMessageToDatabase).toHaveBeenCalledTimes(2);
+
+      // Check the order of calls
+      const calls = TestClient.saveMessageToDatabase.mock.calls;
+      expect(calls[0][0].isCreatedByUser).toBe(true); // First call should be for user message
+      expect(calls[1][0].isCreatedByUser).toBe(false); // Second call should be for response message
+    });
  });
 });
--- a/api/app/clients/specs/FakeClient.js
+++ b/api/app/clients/specs/FakeClient.js
@@ -1,5 +1,5 @@
 const BaseClient = require('../BaseClient');
-const { maxTokensMap } = require('../../../utils');
+const { getModelMaxTokens } = require('../../../utils');

 class FakeClient extends BaseClient {
  constructor(apiKey, options = {}) {
@@ -40,9 +40,9 @@ class FakeClient extends BaseClient {
      };
    }

-    this.maxContextTokens = maxTokensMap[this.modelOptions.model] ?? 4097;
+    this.maxContextTokens =
+      this.options.maxContextTokens ?? getModelMaxTokens(this.modelOptions.model) ?? 4097;
  }
-  getCompletion() {}
  buildMessages() {}
  getTokenCount(str) {
    return str.length;
@@ -65,10 +65,10 @@ const initializeFakeClient = (apiKey, options, fakeMessages) => {
        return Promise.resolve([]);
      }

-      const orderedMessages = TestClient.constructor.getMessagesForConversation(
-        fakeMessages,
+      const orderedMessages = TestClient.constructor.getMessagesForConversation({
+        messages: fakeMessages,
        parentMessageId,
-      );
+      });

      TestClient.currentMessages = orderedMessages;
      return Promise.resolve(orderedMessages);
@@ -86,11 +86,24 @@ 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: [
+        {
+          message: {
+            content: 'Mock response text',
+          },
+        },
+      ],
+    };
+  });
+
  TestClient.buildMessages = jest.fn(async (messages, parentMessageId) => {
-    const orderedMessages = TestClient.constructor.getMessagesForConversation(
+    const orderedMessages = TestClient.constructor.getMessagesForConversation({
      messages,
      parentMessageId,
-    );
+    });
    const formattedMessages = orderedMessages.map((message) => {
      let { role: _role, sender, text } = message;
      const role = _role ?? sender;
--- a/api/app/clients/specs/OpenAIClient.test.js
+++ b/api/app/clients/specs/OpenAIClient.test.js
@@ -1,7 +1,138 @@
+require('dotenv').config();
+const OpenAI = require('openai');
+const { fetchEventSource } = require('@waylaidwanderer/fetch-event-source');
+const { genAzureChatCompletion } = require('~/utils/azureUtils');
 const OpenAIClient = require('../OpenAIClient');
-
 jest.mock('meilisearch');

+jest.mock('~/lib/db/connectDb');
+jest.mock('~/models', () => ({
+  User: jest.fn(),
+  Key: jest.fn(),
+  Session: jest.fn(),
+  Balance: jest.fn(),
+  Transaction: jest.fn(),
+  getMessages: jest.fn().mockResolvedValue([]),
+  saveMessage: jest.fn(),
+  updateMessage: jest.fn(),
+  deleteMessagesSince: jest.fn(),
+  deleteMessages: jest.fn(),
+  getConvoTitle: jest.fn(),
+  getConvo: jest.fn(),
+  saveConvo: jest.fn(),
+  deleteConvos: jest.fn(),
+  getPreset: jest.fn(),
+  getPresets: jest.fn(),
+  savePreset: jest.fn(),
+  deletePresets: jest.fn(),
+  findFileById: jest.fn(),
+  createFile: jest.fn(),
+  updateFile: jest.fn(),
+  deleteFile: jest.fn(),
+  deleteFiles: jest.fn(),
+  getFiles: jest.fn(),
+  updateFileUsage: jest.fn(),
+}));
+
+jest.mock('langchain/chat_models/openai', () => {
+  return {
+    ChatOpenAI: 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);
+});
+
+const finalChatCompletion = jest.fn().mockResolvedValue({
+  choices: [
+    {
+      message: { role: 'assistant', content: 'Mock message content' },
+      finish_reason: 'Mock finish reason',
+    },
+  ],
+});
+
+const stream = jest.fn().mockImplementation(() => {
+  let isDone = false;
+  let isError = false;
+  let errorCallback = null;
+
+  const onEventHandlers = {
+    abort: () => {
+      // Mock abort behavior
+    },
+    error: (callback) => {
+      errorCallback = callback; // Save the error callback for later use
+    },
+    finalMessage: (callback) => {
+      callback({ role: 'assistant', content: 'Mock Response' });
+      isDone = true; // Set stream to done
+    },
+  };
+
+  const mockStream = {
+    on: jest.fn((event, callback) => {
+      if (onEventHandlers[event]) {
+        onEventHandlers[event](callback);
+      }
+      return mockStream;
+    }),
+    finalChatCompletion,
+    controller: { abort: jest.fn() },
+    triggerError: () => {
+      isError = true;
+      if (errorCallback) {
+        errorCallback(new Error('Mock error'));
+      }
+    },
+    [Symbol.asyncIterator]: () => {
+      return {
+        next: () => {
+          if (isError) {
+            return Promise.reject(new Error('Mock error'));
+          }
+          if (isDone) {
+            return Promise.resolve({ done: true });
+          }
+          const chunk = { choices: [{ delta: { content: 'Mock chunk' } }] };
+          return Promise.resolve({ value: chunk, done: false });
+        },
+      };
+    },
+  };
+  return mockStream;
+});
+
+const create = jest.fn().mockResolvedValue({
+  choices: [
+    {
+      message: { content: 'Mock message content' },
+      finish_reason: 'Mock finish reason',
+    },
+  ],
+});
+
+OpenAI.mockImplementation(() => ({
+  beta: {
+    chat: {
+      completions: {
+        stream,
+      },
+    },
+  },
+  chat: {
+    completions: {
+      create,
+    },
+  },
+}));
+
 describe('OpenAIClient', () => {
  let client, client2;
  const model = 'gpt-4';
@@ -11,18 +142,42 @@ describe('OpenAIClient', () => {
    { role: 'assistant', sender: 'Assistant', text: 'Hi', messageId: '2' },
  ];

+  const defaultOptions = {
+    // debug: true,
+    req: {},
+    openaiApiKey: 'new-api-key',
+    modelOptions: {
+      model,
+      temperature: 0.7,
+    },
+  };
+
+  const defaultAzureOptions = {
+    azureOpenAIApiInstanceName: 'your-instance-name',
+    azureOpenAIApiDeploymentName: 'your-deployment-name',
+    azureOpenAIApiVersion: '2020-07-01-preview',
+  };
+
+  let originalWarn;
+
+  beforeAll(() => {
+    originalWarn = console.warn;
+    console.warn = jest.fn();
+  });
+
+  afterAll(() => {
+    console.warn = originalWarn;
+  });
+
  beforeEach(() => {
-    const options = {
-      // debug: true,
-      openaiApiKey: 'new-api-key',
-      modelOptions: {
-        model,
-        temperature: 0.7,
-      },
-    };
+    console.warn.mockClear();
+  });
+
+  beforeEach(() => {
+    const options = { ...defaultOptions };
    client = new OpenAIClient('test-api-key', options);
    client2 = new OpenAIClient('test-api-key', options);
-    client.refineMessages = jest.fn().mockResolvedValue({
+    client.summarizeMessages = jest.fn().mockResolvedValue({
      role: 'assistant',
      content: 'Refined answer',
      tokenCount: 30,
@@ -31,6 +186,7 @@ describe('OpenAIClient', () => {
      .fn()
      .mockResolvedValue({ prompt: messages.map((m) => m.text).join('\n') });
    client.constructor.freeAndResetAllEncoders();
+    client.getMessages = jest.fn().mockResolvedValue([]);
  });

  describe('setOptions', () => {
@@ -39,6 +195,144 @@ describe('OpenAIClient', () => {
      expect(client.modelOptions.model).toBe(model);
      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({});
+      expect(client.FORCE_PROMPT).toBe(true);
+      delete process.env.OPENAI_FORCE_PROMPT; // Cleanup
+      client.FORCE_PROMPT = undefined;
+
+      client.setOptions({ reverseProxyUrl: 'https://example.com/completions' });
+      expect(client.FORCE_PROMPT).toBe(true);
+      client.FORCE_PROMPT = undefined;
+
+      client.setOptions({ reverseProxyUrl: 'https://example.com/chat' });
+      expect(client.FORCE_PROMPT).toBe(false);
+    });
+
+    it('should set isChatCompletion based on useOpenRouter, reverseProxyUrl, or model', () => {
+      client.setOptions({ reverseProxyUrl: null });
+      // true by default since default model will be gpt-3.5-turbo
+      expect(client.isChatCompletion).toBe(true);
+      client.isChatCompletion = undefined;
+
+      // false because completions url will force prompt payload
+      client.setOptions({ reverseProxyUrl: 'https://example.com/completions' });
+      expect(client.isChatCompletion).toBe(false);
+      client.isChatCompletion = undefined;
+
+      client.setOptions({ modelOptions: { model: 'gpt-3.5-turbo' }, reverseProxyUrl: null });
+      expect(client.isChatCompletion).toBe(true);
+    });
+
+    it('should set completionsUrl and langchainProxy based on reverseProxyUrl', () => {
+      client.setOptions({ reverseProxyUrl: 'https://localhost:8080/v1/chat/completions' });
+      expect(client.completionsUrl).toBe('https://localhost:8080/v1/chat/completions');
+      expect(client.langchainProxy).toBe('https://localhost:8080/v1');
+
+      client.setOptions({ reverseProxyUrl: 'https://example.com/completions' });
+      expect(client.completionsUrl).toBe('https://example.com/completions');
+      expect(client.langchainProxy).toBe('https://example.com/completions');
+    });
+  });
+
+  describe('setOptions with Simplified Azure Integration', () => {
+    afterEach(() => {
+      delete process.env.AZURE_OPENAI_DEFAULT_MODEL;
+      delete process.env.AZURE_USE_MODEL_AS_DEPLOYMENT_NAME;
+    });
+
+    const azureOpenAIApiInstanceName = 'test-instance';
+    const azureOpenAIApiDeploymentName = 'test-deployment';
+    const azureOpenAIApiVersion = '2020-07-01-preview';
+
+    const createOptions = (model) => ({
+      modelOptions: { model },
+      azure: {
+        azureOpenAIApiInstanceName,
+        azureOpenAIApiDeploymentName,
+        azureOpenAIApiVersion,
+      },
+    });
+
+    it('should set model from AZURE_OPENAI_DEFAULT_MODEL when Azure is enabled', () => {
+      process.env.AZURE_OPENAI_DEFAULT_MODEL = 'gpt-4-azure';
+      const options = createOptions('test');
+      client.azure = options.azure;
+      client.setOptions(options);
+      expect(client.modelOptions.model).toBe('gpt-4-azure');
+    });
+
+    it('should not change model if Azure is not enabled', () => {
+      process.env.AZURE_OPENAI_DEFAULT_MODEL = 'gpt-4-azure';
+      const originalModel = 'test';
+      client.azure = false;
+      client.setOptions(createOptions('test'));
+      expect(client.modelOptions.model).toBe(originalModel);
+    });
+
+    it('should not change model if AZURE_OPENAI_DEFAULT_MODEL is not set and model is passed', () => {
+      const originalModel = 'GROK-LLM';
+      const options = createOptions(originalModel);
+      client.azure = options.azure;
+      client.setOptions(options);
+      expect(client.modelOptions.model).toBe(originalModel);
+    });
+
+    it('should change model if AZURE_OPENAI_DEFAULT_MODEL is set and model is passed', () => {
+      process.env.AZURE_OPENAI_DEFAULT_MODEL = 'gpt-4-azure';
+      const originalModel = 'GROK-LLM';
+      const options = createOptions(originalModel);
+      client.azure = options.azure;
+      client.setOptions(options);
+      expect(client.modelOptions.model).toBe(process.env.AZURE_OPENAI_DEFAULT_MODEL);
+    });
+
+    it('should include model in deployment name if AZURE_USE_MODEL_AS_DEPLOYMENT_NAME is set', () => {
+      process.env.AZURE_USE_MODEL_AS_DEPLOYMENT_NAME = 'true';
+      const model = 'gpt-4-azure';
+
+      const AzureClient = new OpenAIClient('test-api-key', createOptions(model));
+
+      const expectedValue = `https://${azureOpenAIApiInstanceName}.openai.azure.com/openai/deployments/${model}/chat/completions?api-version=${azureOpenAIApiVersion}`;
+
+      expect(AzureClient.modelOptions.model).toBe(model);
+      expect(AzureClient.azureEndpoint).toBe(expectedValue);
+    });
+
+    it('should include model in deployment name if AZURE_USE_MODEL_AS_DEPLOYMENT_NAME and default model is set', () => {
+      const defaultModel = 'gpt-4-azure';
+      process.env.AZURE_USE_MODEL_AS_DEPLOYMENT_NAME = 'true';
+      process.env.AZURE_OPENAI_DEFAULT_MODEL = defaultModel;
+      const model = 'gpt-4-this-is-a-test-model-name';
+
+      const AzureClient = new OpenAIClient('test-api-key', createOptions(model));
+
+      const expectedValue = `https://${azureOpenAIApiInstanceName}.openai.azure.com/openai/deployments/${model}/chat/completions?api-version=${azureOpenAIApiVersion}`;
+
+      expect(AzureClient.modelOptions.model).toBe(defaultModel);
+      expect(AzureClient.azureEndpoint).toBe(expectedValue);
+    });
+
+    it('should not include model in deployment name if AZURE_USE_MODEL_AS_DEPLOYMENT_NAME is not set', () => {
+      const model = 'gpt-4-azure';
+
+      const AzureClient = new OpenAIClient('test-api-key', createOptions(model));
+
+      const expectedValue = `https://${azureOpenAIApiInstanceName}.openai.azure.com/openai/deployments/${azureOpenAIApiDeploymentName}/chat/completions?api-version=${azureOpenAIApiVersion}`;
+
+      expect(AzureClient.modelOptions.model).toBe(model);
+      expect(AzureClient.azureEndpoint).toBe(expectedValue);
+    });
  });

  describe('selectTokenizer', () => {
@@ -158,7 +452,7 @@ describe('OpenAIClient', () => {
    });

    it('should handle context strategy correctly', async () => {
-      client.contextStrategy = 'refine';
+      client.contextStrategy = 'summarize';
      const result = await client.buildMessages(messages, parentMessageId, {
        isChatCompletion: true,
      });
@@ -172,22 +466,11 @@ describe('OpenAIClient', () => {
        isChatCompletion: true,
      });
      const hasUserWithName = result.prompt.some(
-        (item) => item.role === 'user' && item.name === 'Test User',
+        (item) => item.role === 'user' && item.name === 'Test_User',
      );
      expect(hasUserWithName).toBe(true);
    });

-    it('should calculate tokenCount for each message when contextStrategy is set', async () => {
-      client.contextStrategy = 'refine';
-      const result = await client.buildMessages(messages, parentMessageId, {
-        isChatCompletion: true,
-      });
-      const hasUserWithTokenCount = result.prompt.some(
-        (item) => item.role === 'user' && item.tokenCount > 0,
-      );
-      expect(hasUserWithTokenCount).toBe(true);
-    });
-
    it('should handle promptPrefix from options when promptPrefix argument is not provided', async () => {
      client.options.promptPrefix = 'Test Prefix from options';
      const result = await client.buildMessages(messages, parentMessageId, {
@@ -271,5 +554,151 @@ describe('OpenAIClient', () => {
        expect(totalTokens).toBe(testCase.expected);
      });
    });
+
+    const vision_request = [
+      {
+        role: 'user',
+        content: [
+          {
+            type: 'text',
+            text: 'describe what is in this image?',
+          },
+          {
+            type: 'image_url',
+            image_url: {
+              url: 'https://venturebeat.com/wp-content/uploads/2019/03/openai-1.png',
+              detail: 'high',
+            },
+          },
+        ],
+      },
+    ];
+
+    const expectedTokens = 14;
+    const visionModel = 'gpt-4-vision-preview';
+
+    it(`should return ${expectedTokens} tokens for model ${visionModel} (Vision Request)`, () => {
+      client.modelOptions.model = visionModel;
+      client.selectTokenizer();
+      // 3 tokens for assistant label
+      let totalTokens = 3;
+      for (let message of vision_request) {
+        totalTokens += client.getTokenCountForMessage(message);
+      }
+      expect(totalTokens).toBe(expectedTokens);
+    });
+  });
+
+  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);
+
+      const currentDateString = new Date().toLocaleDateString('en-us', {
+        year: 'numeric',
+        month: 'long',
+        day: 'numeric',
+      });
+
+      expect(getCompletion.mock.calls[0][0]).toBe(
+        `||>Instructions:\nYou are ChatGPT, a large language model trained by OpenAI. Respond conversationally.\nCurrent date: ${currentDateString}\n\n||>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' }];
+
+    beforeEach(() => {
+      client = new OpenAIClient('test-api-key', {
+        endpoint: 'ollama',
+        modelOptions: {
+          model: 'initial-model',
+        },
+        modelsConfig: {
+          ollama: ['initial-model', 'llava', 'other-model'],
+        },
+      });
+
+      client.defaultVisionModel = 'non-valid-default-model';
+    });
+
+    afterEach(() => {
+      jest.restoreAllMocks();
+    });
+
+    it('should set "llava" as the model if it is the first valid model when default validation fails', () => {
+      client.checkVisionRequest(attachments);
+
+      expect(client.modelOptions.model).toBe('llava');
+      expect(client.isVisionModel).toBeTruthy();
+      expect(client.modelOptions.stop).toBeUndefined();
+    });
  });
 });
--- a/api/app/clients/specs/OpenAIClient.tokens.js
+++ b/api/app/clients/specs/OpenAIClient.tokens.js
@@ -38,7 +38,12 @@ const run = async () => {
  "On the other hand, we denounce with righteous indignation and dislike men who are so beguiled and demoralized by the charms of pleasure of the moment, so blinded by desire, that they cannot foresee the pain and trouble that are bound to ensue; and equal blame belongs to those who fail in their duty through weakness of will, which is the same as saying through shrinking from toil and pain. These cases are perfectly simple and easy to distinguish. In a free hour, when our power of choice is untrammelled and when nothing prevents our being able to do what we like best, every pleasure is to be welcomed and every pain avoided. But in certain circumstances and owing to the claims of duty or the obligations of business it will frequently occur that pleasures have to be repudiated and annoyances accepted. The wise man therefore always holds in these matters to this principle of selection: he rejects pleasures to secure other greater pleasures, or else he endures pains to avoid worse pains."
  `;
  const model = 'gpt-3.5-turbo';
-  const maxContextTokens = model === 'gpt-4' ? 8191 : model === 'gpt-4-32k' ? 32767 : 4095; // 1 less than maximum
+  let maxContextTokens = 4095;
+  if (model === 'gpt-4') {
+    maxContextTokens = 8191;
+  } else if (model === 'gpt-4-32k') {
+    maxContextTokens = 32767;
+  }
  const clientOptions = {
    reverseProxyUrl: process.env.OPENAI_REVERSE_PROXY || null,
    maxContextTokens,
--- a/api/app/clients/specs/PluginsClient.test.js
+++ b/api/app/clients/specs/PluginsClient.test.js
@@ -1,9 +1,10 @@
+const crypto = require('crypto');
+const { Constants } = require('librechat-data-provider');
 const { HumanChatMessage, AIChatMessage } = require('langchain/schema');
 const PluginsClient = require('../PluginsClient');
-const crypto = require('crypto');

-jest.mock('../../../lib/db/connectDb');
-jest.mock('../../../models/Conversation', () => {
+jest.mock('~/lib/db/connectDb');
+jest.mock('~/models/Conversation', () => {
  return function () {
    return {
      save: jest.fn(),
@@ -12,6 +13,12 @@ jest.mock('../../../models/Conversation', () => {
  };
 });

+const defaultAzureOptions = {
+  azureOpenAIApiInstanceName: 'your-instance-name',
+  azureOpenAIApiDeploymentName: 'your-deployment-name',
+  azureOpenAIApiVersion: '2020-07-01-preview',
+};
+
 describe('PluginsClient', () => {
  let TestAgent;
  let options = {
@@ -41,10 +48,10 @@ describe('PluginsClient', () => {
          return Promise.resolve([]);
        }

-        const orderedMessages = TestAgent.constructor.getMessagesForConversation(
-          fakeMessages,
+        const orderedMessages = TestAgent.constructor.getMessagesForConversation({
+          messages: fakeMessages,
          parentMessageId,
-        );
+        });

        const chatMessages = orderedMessages.map((msg) =>
          msg?.isCreatedByUser || msg?.role?.toLowerCase() === 'user'
@@ -60,7 +67,7 @@ describe('PluginsClient', () => {
        TestAgent.setOptions(opts);
      }
      const conversationId = opts.conversationId || crypto.randomUUID();
-      const parentMessageId = opts.parentMessageId || '00000000-0000-0000-0000-000000000000';
+      const parentMessageId = opts.parentMessageId || Constants.NO_PARENT;
      const userMessageId = opts.overrideParentMessageId || crypto.randomUUID();
      this.pastMessages = await TestAgent.loadHistory(
        conversationId,
@@ -144,4 +151,164 @@ describe('PluginsClient', () => {
      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/.well-known/Diagrams.json
+++ b/api/app/clients/tools/.well-known/Diagrams.json
--- a/api/app/clients/tools/.well-known/aitoolhunt.json
+++ b/api/app/clients/tools/.well-known/aitoolhunt.json
@@ -1,22 +0,0 @@
-{
-  "schema_version": "v1",
-  "name_for_human": "Ai Tool Hunt",
-  "name_for_model": "aitoolhunt",
-  "description_for_human": "Find the perfect AI tools for all your needs, drawn from the most comprehensive global database of AI tools.",
-  "description_for_model": "This universal plugin is designed to identify and suggest AI tools for any imaginable task or challenge. It harnesses a database of over 1600+ AI tools and 120+ categories, ensuring a solution for every requirement. Simply input any task, problem, or use case, and the plugin will generate a tailored list of AI tools. Note: it's crucial to rearrange the tools suggested by the plugin based on their relevance to the user's specific use case. This plugin can tackle everything from the simplest to the most intricate tasks. If there's a way an AI tool can help, this plugin will not only locate it but also customize the list to the user's unique needs. Utilize this plugin whenever a task or activity emerges, regardless of its context or complexity. When formulating a search query, avoid stop words or other irrelevant keywords. For instance, 'copywriting' is acceptable, but 'ai for copywriting' is not. If you believe none of the suggested tools are a suitable match for the user's needs, indicate that these are related tools.",
-  "auth": {
-    "type": "service_http",
-    "authorization_type": "bearer",
-    "verification_tokens": {
-      "openai": "06a0f9391a5e48c7a7eeaca1e7e1e8d3"
-    }
-  },
-  "api": {
-    "type": "openapi",
-    "url": "https://www.aitoolhunt.com/openapi.json",
-    "is_user_authenticated": false
-  },
-  "logo_url": "https://www.aitoolhunt.com/images/aitoolhunt_logo.png",
-  "contact_email": "aitoolhunt@gmail.com",
-  "legal_info_url": "https://www.aitoolhunt.com/terms-and-conditions"
-}
--- a/api/app/clients/tools/.well-known/rephrase.json
+++ b/api/app/clients/tools/.well-known/rephrase.json
@@ -1,18 +0,0 @@
-{
-  "schema_version": "v1",
-  "name_for_human": "Prompt Perfect",
-  "name_for_model": "rephrase",
-  "description_for_human": "Type 'perfect' to craft the perfect prompt, every time.",
-  "description_for_model": "Plugin that can rephrase user inputs to improve the quality of ChatGPT's responses. The plugin evaluates user inputs and, if necessary, transforms them into clearer, more specific, and contextual prompts. It processes a JSON object containing the user input to be rephrased and uses the GPT-3.5-turbo model for the rephrasing process. The rephrased input is then returned as raw data to be incorporated into ChatGPT's response. The user can initiate the plugin by typing 'perfect'.",
-  "auth": {
-    "type": "none"
-  },
-  "api": {
-    "type": "openapi",
-    "url": "https://promptperfect.xyz/openapi.yaml",
-    "is_user_authenticated": false
-  },
-  "logo_url": "https://promptperfect.xyz/static/prompt_perfect_logo.png",
-  "contact_email": "heyo@promptperfect.xyz",
-  "legal_info_url": "https://promptperfect.xyz/static/terms.html"
-}
--- a/api/app/clients/tools/AIPluginTool.js
+++ b/api/app/clients/tools/AIPluginTool.js
@@ -1,238 +0,0 @@
-const { Tool } = require('langchain/tools');
-const yaml = require('js-yaml');
-
-/*
-export interface AIPluginToolParams {
-  name: string;
-  description: string;
-  apiSpec: string;
-  openaiSpec: string;
-  model: BaseLanguageModel;
-}
-
-export interface PathParameter {
-  name: string;
-  description: string;
-}
-
-export interface Info {
-  title: string;
-  description: string;
-  version: string;
-}
-export interface PathMethod {
-  summary: string;
-  operationId: string;
-  parameters?: PathParameter[];
-}
-
-interface ApiSpec {
-  openapi: string;
-  info: Info;
-  paths: { [key: string]: { [key: string]: PathMethod } };
-}
-*/
-
-function isJson(str) {
-  try {
-    JSON.parse(str);
-  } catch (e) {
-    return false;
-  }
-  return true;
-}
-
-function convertJsonToYamlIfApplicable(spec) {
-  if (isJson(spec)) {
-    const jsonData = JSON.parse(spec);
-    return yaml.dump(jsonData);
-  }
-  return spec;
-}
-
-function extractShortVersion(openapiSpec) {
-  openapiSpec = convertJsonToYamlIfApplicable(openapiSpec);
-  try {
-    const fullApiSpec = yaml.load(openapiSpec);
-    const shortApiSpec = {
-      openapi: fullApiSpec.openapi,
-      info: fullApiSpec.info,
-      paths: {},
-    };
-
-    for (let path in fullApiSpec.paths) {
-      shortApiSpec.paths[path] = {};
-      for (let method in fullApiSpec.paths[path]) {
-        shortApiSpec.paths[path][method] = {
-          summary: fullApiSpec.paths[path][method].summary,
-          operationId: fullApiSpec.paths[path][method].operationId,
-          parameters: fullApiSpec.paths[path][method].parameters?.map((parameter) => ({
-            name: parameter.name,
-            description: parameter.description,
-          })),
-        };
-      }
-    }
-
-    return yaml.dump(shortApiSpec);
-  } catch (e) {
-    console.log(e);
-    return '';
-  }
-}
-function printOperationDetails(operationId, openapiSpec) {
-  openapiSpec = convertJsonToYamlIfApplicable(openapiSpec);
-  let returnText = '';
-  try {
-    let doc = yaml.load(openapiSpec);
-    let servers = doc.servers;
-    let paths = doc.paths;
-    let components = doc.components;
-
-    for (let path in paths) {
-      for (let method in paths[path]) {
-        let operation = paths[path][method];
-        if (operation.operationId === operationId) {
-          returnText += `The API request to do for operationId "${operationId}" is:\n`;
-          returnText += `Method: ${method.toUpperCase()}\n`;
-
-          let url = servers[0].url + path;
-          returnText += `Path: ${url}\n`;
-
-          returnText += 'Parameters:\n';
-          if (operation.parameters) {
-            for (let param of operation.parameters) {
-              let required = param.required ? '' : ' (optional),';
-              returnText += `- ${param.name} (${param.in},${required} ${param.schema.type}): ${param.description}\n`;
-            }
-          } else {
-            returnText += ' None\n';
-          }
-          returnText += '\n';
-
-          let responseSchema = operation.responses['200'].content['application/json'].schema;
-
-          // Check if schema is a reference
-          if (responseSchema.$ref) {
-            // Extract schema name from reference
-            let schemaName = responseSchema.$ref.split('/').pop();
-            // Look up schema in components
-            responseSchema = components.schemas[schemaName];
-          }
-
-          returnText += 'Response schema:\n';
-          returnText += '- Type: ' + responseSchema.type + '\n';
-          returnText += '- Additional properties:\n';
-          returnText += '  - Type: ' + responseSchema.additionalProperties?.type + '\n';
-          if (responseSchema.additionalProperties?.properties) {
-            returnText += '  - Properties:\n';
-            for (let prop in responseSchema.additionalProperties.properties) {
-              returnText += `    - ${prop} (${responseSchema.additionalProperties.properties[prop].type}): Description not provided in OpenAPI spec\n`;
-            }
-          }
-        }
-      }
-    }
-    if (returnText === '') {
-      returnText += `No operation with operationId "${operationId}" found.`;
-    }
-    return returnText;
-  } catch (e) {
-    console.log(e);
-    return '';
-  }
-}
-
-class AIPluginTool extends Tool {
-  /*
-  private _name: string;
-  private _description: string;
-  apiSpec: string;
-  openaiSpec: string;
-  model: BaseLanguageModel;
-  */
-
-  get name() {
-    return this._name;
-  }
-
-  get description() {
-    return this._description;
-  }
-
-  constructor(params) {
-    super();
-    this._name = params.name;
-    this._description = params.description;
-    this.apiSpec = params.apiSpec;
-    this.openaiSpec = params.openaiSpec;
-    this.model = params.model;
-  }
-
-  async _call(input) {
-    let date = new Date();
-    let fullDate = `Date: ${date.getDate()}/${
-      date.getMonth() + 1
-    }/${date.getFullYear()}, Time: ${date.getHours()}:${date.getMinutes()}:${date.getSeconds()}`;
-    const prompt = `${fullDate}\nQuestion: ${input} \n${this.apiSpec}.`;
-    console.log(prompt);
-    const gptResponse = await this.model.predict(prompt);
-    let operationId = gptResponse.match(/operationId: (.*)/)?.[1];
-    if (!operationId) {
-      return 'No operationId found in the response';
-    }
-    if (operationId == 'No API path found to answer the question') {
-      return 'No API path found to answer the question';
-    }
-
-    let openApiData = printOperationDetails(operationId, this.openaiSpec);
-
-    return openApiData;
-  }
-
-  static async fromPluginUrl(url, model) {
-    const aiPluginRes = await fetch(url, {});
-    if (!aiPluginRes.ok) {
-      throw new Error(`Failed to fetch plugin from ${url} with status ${aiPluginRes.status}`);
-    }
-    const aiPluginJson = await aiPluginRes.json();
-    const apiUrlRes = await fetch(aiPluginJson.api.url, {});
-    if (!apiUrlRes.ok) {
-      throw new Error(
-        `Failed to fetch API spec from ${aiPluginJson.api.url} with status ${apiUrlRes.status}`,
-      );
-    }
-    const apiUrlJson = await apiUrlRes.text();
-    const shortApiSpec = extractShortVersion(apiUrlJson);
-    return new AIPluginTool({
-      name: aiPluginJson.name_for_model.toLowerCase(),
-      description: `A \`tool\` to learn the API documentation for ${aiPluginJson.name_for_model.toLowerCase()}, after which you can use 'http_request' to make the actual API call. Short description of how to use the API's results: ${
-        aiPluginJson.description_for_model
-      })`,
-      apiSpec: `
-As an AI, your task is to identify the operationId of the relevant API path based on the condensed OpenAPI specifications provided.
-
-Please note:
-
-1. Do not imagine URLs. Only use the information provided in the condensed OpenAPI specifications.
-
-2. Do not guess the operationId. Identify it strictly based on the API paths and their descriptions.
-
-Your output should only include:
- operationId: The operationId of the relevant API path
-
-If you cannot find a suitable API path based on the OpenAPI specifications, please answer only "operationId: No API path found to answer the question".
-
-Now, based on the question above and the condensed OpenAPI specifications given below, identify the operationId:
-
-\`\`\`
-${shortApiSpec}
-\`\`\`
-`,
-      openaiSpec: apiUrlJson,
-      model: model,
-    });
-  }
-}
-
-module.exports = AIPluginTool;
--- a/api/app/clients/tools/AzureAiSearch.js
+++ b/api/app/clients/tools/AzureAiSearch.js
@@ -0,0 +1,98 @@
+const { z } = require('zod');
+const { StructuredTool } = require('langchain/tools');
+const { SearchClient, AzureKeyCredential } = require('@azure/search-documents');
+const { logger } = require('~/config');
+
+class AzureAISearch extends StructuredTool {
+  // Constants for default values
+  static DEFAULT_API_VERSION = '2023-11-01';
+  static DEFAULT_QUERY_TYPE = 'simple';
+  static DEFAULT_TOP = 5;
+
+  // Helper function for initializing properties
+  _initializeField(field, envVar, defaultValue) {
+    return field || process.env[envVar] || defaultValue;
+  }
+
+  constructor(fields = {}) {
+    super();
+    this.name = 'azure-ai-search';
+    this.description =
+      'Use the \'azure-ai-search\' tool to retrieve search results relevant to your input';
+
+    // Initialize properties using helper function
+    this.serviceEndpoint = this._initializeField(
+      fields.AZURE_AI_SEARCH_SERVICE_ENDPOINT,
+      'AZURE_AI_SEARCH_SERVICE_ENDPOINT',
+    );
+    this.indexName = this._initializeField(
+      fields.AZURE_AI_SEARCH_INDEX_NAME,
+      'AZURE_AI_SEARCH_INDEX_NAME',
+    );
+    this.apiKey = this._initializeField(fields.AZURE_AI_SEARCH_API_KEY, 'AZURE_AI_SEARCH_API_KEY');
+    this.apiVersion = this._initializeField(
+      fields.AZURE_AI_SEARCH_API_VERSION,
+      'AZURE_AI_SEARCH_API_VERSION',
+      AzureAISearch.DEFAULT_API_VERSION,
+    );
+    this.queryType = this._initializeField(
+      fields.AZURE_AI_SEARCH_SEARCH_OPTION_QUERY_TYPE,
+      'AZURE_AI_SEARCH_SEARCH_OPTION_QUERY_TYPE',
+      AzureAISearch.DEFAULT_QUERY_TYPE,
+    );
+    this.top = this._initializeField(
+      fields.AZURE_AI_SEARCH_SEARCH_OPTION_TOP,
+      'AZURE_AI_SEARCH_SEARCH_OPTION_TOP',
+      AzureAISearch.DEFAULT_TOP,
+    );
+    this.select = this._initializeField(
+      fields.AZURE_AI_SEARCH_SEARCH_OPTION_SELECT,
+      'AZURE_AI_SEARCH_SEARCH_OPTION_SELECT',
+    );
+
+    // Check for required fields
+    if (!this.serviceEndpoint || !this.indexName || !this.apiKey) {
+      throw new Error(
+        'Missing AZURE_AI_SEARCH_SERVICE_ENDPOINT, AZURE_AI_SEARCH_INDEX_NAME, or AZURE_AI_SEARCH_API_KEY environment variable.',
+      );
+    }
+
+    // Create SearchClient
+    this.client = new SearchClient(
+      this.serviceEndpoint,
+      this.indexName,
+      new AzureKeyCredential(this.apiKey),
+      { apiVersion: this.apiVersion },
+    );
+
+    // Define schema
+    this.schema = z.object({
+      query: z.string().describe('Search word or phrase to Azure AI Search'),
+    });
+  }
+
+  // Improved error handling and logging
+  async _call(data) {
+    const { query } = data;
+    try {
+      const searchOption = {
+        queryType: this.queryType,
+        top: this.top,
+      };
+      if (this.select) {
+        searchOption.select = this.select.split(',');
+      }
+      const searchResults = await this.client.search(query, searchOption);
+      const resultDocuments = [];
+      for await (const result of searchResults.results) {
+        resultDocuments.push(result.document);
+      }
+      return JSON.stringify(resultDocuments);
+    } catch (error) {
+      logger.error('Azure AI Search request failed', error);
+      return 'There was an error with Azure AI Search.';
+    }
+  }
+}
+
+module.exports = AzureAISearch;
--- a/api/app/clients/tools/AzureCognitiveSearch.js
+++ b/api/app/clients/tools/AzureCognitiveSearch.js
@@ -1,111 +0,0 @@
-const { Tool } = require('langchain/tools');
-const { SearchClient, AzureKeyCredential } = require('@azure/search-documents');
-
-class AzureCognitiveSearch extends Tool {
-  constructor(fields = {}) {
-    super();
-    this.serviceEndpoint =
-      fields.AZURE_COGNITIVE_SEARCH_SERVICE_ENDPOINT || this.getServiceEndpoint();
-    this.indexName = fields.AZURE_COGNITIVE_SEARCH_INDEX_NAME || this.getIndexName();
-    this.apiKey = fields.AZURE_COGNITIVE_SEARCH_API_KEY || this.getApiKey();
-
-    this.apiVersion = fields.AZURE_COGNITIVE_SEARCH_API_VERSION || this.getApiVersion();
-
-    this.queryType = fields.AZURE_COGNITIVE_SEARCH_SEARCH_OPTION_QUERY_TYPE || this.getQueryType();
-    this.top = fields.AZURE_COGNITIVE_SEARCH_SEARCH_OPTION_TOP || this.getTop();
-    this.select = fields.AZURE_COGNITIVE_SEARCH_SEARCH_OPTION_SELECT || this.getSelect();
-
-    this.client = new SearchClient(
-      this.serviceEndpoint,
-      this.indexName,
-      new AzureKeyCredential(this.apiKey),
-      {
-        apiVersion: this.apiVersion,
-      },
-    );
-  }
-
-  /**
-   * The name of the tool.
-   * @type {string}
-   */
-  name = 'azure-cognitive-search';
-
-  /**
-   * A description for the agent to use
-   * @type {string}
-   */
-  description =
-    'Use the \'azure-cognitive-search\' tool to retrieve search results relevant to your input';
-
-  getServiceEndpoint() {
-    const serviceEndpoint = process.env.AZURE_COGNITIVE_SEARCH_SERVICE_ENDPOINT || '';
-    if (!serviceEndpoint) {
-      throw new Error('Missing AZURE_COGNITIVE_SEARCH_SERVICE_ENDPOINT environment variable.');
-    }
-    return serviceEndpoint;
-  }
-
-  getIndexName() {
-    const indexName = process.env.AZURE_COGNITIVE_SEARCH_INDEX_NAME || '';
-    if (!indexName) {
-      throw new Error('Missing AZURE_COGNITIVE_SEARCH_INDEX_NAME environment variable.');
-    }
-    return indexName;
-  }
-
-  getApiKey() {
-    const apiKey = process.env.AZURE_COGNITIVE_SEARCH_API_KEY || '';
-    if (!apiKey) {
-      throw new Error('Missing AZURE_COGNITIVE_SEARCH_API_KEY environment variable.');
-    }
-    return apiKey;
-  }
-
-  getApiVersion() {
-    return process.env.AZURE_COGNITIVE_SEARCH_API_VERSION || '2020-06-30';
-  }
-
-  getQueryType() {
-    return process.env.AZURE_COGNITIVE_SEARCH_SEARCH_OPTION_QUERY_TYPE || 'simple';
-  }
-
-  getTop() {
-    if (process.env.AZURE_COGNITIVE_SEARCH_SEARCH_OPTION_TOP) {
-      return Number(process.env.AZURE_COGNITIVE_SEARCH_SEARCH_OPTION_TOP);
-    } else {
-      return 5;
-    }
-  }
-
-  getSelect() {
-    if (process.env.AZURE_COGNITIVE_SEARCH_SEARCH_OPTION_SELECT) {
-      return process.env.AZURE_COGNITIVE_SEARCH_SEARCH_OPTION_SELECT.split(',');
-    } else {
-      return null;
-    }
-  }
-
-  async _call(query) {
-    try {
-      const searchOption = {
-        queryType: this.queryType,
-        top: this.top,
-      };
-      if (this.select) {
-        searchOption.select = this.select;
-      }
-      const searchResults = await this.client.search(query, searchOption);
-      const resultDocuments = [];
-      for await (const result of searchResults.results) {
-        resultDocuments.push(result.document);
-      }
-      return JSON.stringify(resultDocuments);
-    } catch (error) {
-      console.error(`Azure Cognitive Search request failed: ${error}`);
-      return 'There was an error with Azure Cognitive Search.';
-    }
-  }
-}
-
-module.exports = AzureCognitiveSearch;
--- a/api/app/clients/tools/CodeInterpreter.js
+++ b/api/app/clients/tools/CodeInterpreter.js
@@ -1,52 +0,0 @@
-const { Tool } = require('langchain/tools');
-const WebSocket = require('ws');
-const { promisify } = require('util');
-const fs = require('fs');
-
-class CodeInterpreter extends Tool {
-  constructor() {
-    super();
-    this.name = 'code-interpreter';
-    this.description = `If there is plotting or any image related tasks, save the result as .png file. 
-    No need show the image or plot. USE print(variable_name) if you need output.You can run python codes with this plugin.You have to use print function in python code to get any result from this plugin. 
-    This does not support user input. Even if the code has input() function, change it to an appropriate value.
-    You can show the user the code with input() functions. But the code passed to the plug-in should not contain input().
-    You should provide properly formatted code to this plugin. If the code is executed successfully, the stdout will be returned to you. You have to print that to the user, and if the user had 
-    asked for an explanation, you have to provide one. If the output is "Error From here" or any other error message, 
-    tell the user "Python Engine Failed" and continue with whatever you are supposed to do.`;
-
-    // Create a promisified version of fs.unlink
-    this.unlinkAsync = promisify(fs.unlink);
-  }
-
-  async _call(input) {
-    const websocket = new WebSocket('ws://localhost:3380'); // Update with your WebSocket server URL
-
-    // Wait until the WebSocket connection is open
-    await new Promise((resolve) => {
-      websocket.onopen = resolve;
-    });
-
-    // Send the Python code to the server
-    websocket.send(input);
-
-    // Wait for the result from the server
-    const result = await new Promise((resolve) => {
-      websocket.onmessage = (event) => {
-        resolve(event.data);
-      };
-
-      // Handle WebSocket connection closed
-      websocket.onclose = () => {
-        resolve('Python Engine Failed');
-      };
-    });
-
-    // Close the WebSocket connection
-    websocket.close();
-
-    return result;
-  }
-}
-
-module.exports = CodeInterpreter;
--- a/Show More
+++ b/Show More