librechat.n2a.yml

version: 1.0.0
cache: true
interface:
  agents: true
  runCode: false
  modelSelect: true
  parameters: true
  presets: true
  prompts: true
modelSpecs:
  addedEndpoints:
    - agents
    - azureOpenAI
    - "Gemini 2.5 Pro"
    - "Gemini 2.5 Flash"
    - "Claude Sonnet 4.5"
  list:
    - name: azure-gpt-4o
      label: GPT-4o
      default: true
      description: For daily tasks with a quick response time.
      preset:
        endpoint: azureOpenAI
        model: gpt-4o
    - name: azure-gpt-4o-mini
      label: GPT-4o Mini
      description: Offers simpler, more instant responses.
      preset:
        endpoint: azureOpenAI
        model: gpt-4o-mini
    - name: azure-gpt-41
      label: GPT-4.1
      description: For complex reasoning and coding.
      preset:
        endpoint: azureOpenAI
        model: gpt-41
    - name: azure-gpt-41-mini
      label: GPT-4.1 Mini
      description: A faster, more lightweight version of GPT-4.1.
      preset:
        endpoint: azureOpenAI
        model: gpt-41-mini
    - name: azure-gpt-5
      label: GPT-5
      description: For advanced tasks with the highest accuracy and reasoning depth.
      preset:
        endpoint: azureOpenAI
        model: gpt-5
    - name: google-gemini-25-pro
      label: Gemini 2.5 Pro
      description: Quickly analyze complex, mixed media content.
      preset:
        endpoint: "Gemini 2.5 Pro"
        model: gemini-2.5-pro
    - name: google-gemini-25-flash
      label: Gemini 2.5 Flash
      description: A faster, more lightweight version of Gemini 2.5 Pro.
      preset:
        endpoint: "Gemini 2.5 Flash"
        model: gemini-2.5-flash
    - name: anthropic-claude-sonnet-45
      label: Claude Sonnet 4.5 (Limited)
      description: For extended reasoning, coding, and automation. This model is being evaluated. It can be used a limited number of times.
      preset:
        endpoint: "Claude Sonnet 4.5"
        model: claude-sonnet-4-5
mcpServers:
  Internet Search (Tavily):
    type: streamable-http
    url: https://mcp.tavily.com/mcp/
    headers:
      Authorization: "Bearer ${TAVILY_API_KEY}"
    startup: true
    chatMenu: false
    serverInstructions: "TAVILY SEARCH PROTOCOL: (1) Never use tavily_research tool unless the user explicitly includes '/deep-research' in their prompt. For all other queries, use tavily_search which has higher rate limits. If /deep-research is requested, use 'mini' model first, and only escalate to 'pro' if the task clearly requires broader multi-topic coverage. (2) Before invoking any Tavily function, validate that the 'query' parameter contains a non-empty string and all parameters match the expected API schema. (3) Execute searches ONE AT A TIME - wait for and verify results from the current search before initiating any additional searches. Never launch multiple parallel search requests. (4) If a search fails or returns errors, analyze the error message, modify the query or parameters, and retry with adjustments. Continue iterating up to 25 attempts to achieve a successful search."
    requiresOAuth: false
endpoints:
  azureOpenAI:
    # Endpoint-level configuration
    titleModel: "gpt-4o"
    plugins: false
    assistants: false
    groups:
    # Group-level configuration
    - group: "N2A East US"
      instanceName: "N2A OpenAI"
      apiKey: "${AZURE_OPENAI_API_KEY}"
      baseURL: "${AZURE_OPENAI_BASEURL}"
      version: "2024-10-21"
      # Model-level configuration
      models:
        gpt-5:
          deploymentName: gpt-5
          version: "2024-10-21"
        gpt-4o:
          deploymentName: gpt-4o
          version: "2024-10-21"
        gpt-4o-mini:
          deploymentName: gpt-4o-mini
          version: "2024-10-21"
        gpt-41:
          deploymentName: gpt-41
          version: "2024-10-21"
        gpt-41-mini:
          deploymentName: gpt-41-mini
          version: "2024-10-21"
  custom:
    - name: "Gemini 2.5 Pro"
      apiKey: "${GCP_VERTEXAI_API_KEY}"
      baseURL: "${GCP_VERTEXAI_BASEURL}/gemini-2.5-pro/v1"
      directEndpoint: true
      models:
        default:
          - gemini-2.5-pro
      titleModel: "gemini-2.5-pro"
      titleConvo: true
      modelDisplayLabel: "Gemini 2.5 Pro"
      addParams:
        streaming: false
        disableStreaming: true
      dropParams:
        - stream
        - streaming
    - name: "Gemini 2.5 Flash"
      apiKey: "${GCP_VERTEXAI_API_KEY}"
      baseURL: "${GCP_VERTEXAI_BASEURL}/gemini-2.5-flash/v1"
      directEndpoint: true
      models:
        default:
          - gemini-2.5-flash
      titleModel: "gemini-2.5-flash"
      titleConvo: true
      modelDisplayLabel: "Gemini 2.5 Flash"
      addParams:
        streaming: false
        disableStreaming: true
      dropParams:
        - stream
        - streaming
    - name: "Claude Sonnet 4.5"
      apiKey: "${GCP_VERTEXAI_API_KEY}"
      baseURL: "${GCP_VERTEXAI_BASEURL}/claude-sonnet-4-5/v1"
      directEndpoint: true
      models:
        default:
          - claude-sonnet-4-5
      titleModel: "claude-sonnet-4-5"
      titleConvo: true
      modelDisplayLabel: "Claude Sonnet 4.5"
      # Streaming enabled - Kong translates to OpenAI SSE format
      streamRate: 25
    - name: "LangGraph Agents"
      apiKey: "${LANGGRAPH_PROXY_API_KEY}"
      baseURL: "${LANGGRAPH_PROXY_BASE_URL}"
      models:
        default:
          - langgraph-agent
        fetch: true
      titleConvo: true
      modelDisplayLabel: "LangGraph Agents"
      streamRate: 25
      dropParams:
        - 'stop'
        - 'frequency_penalty'
        - 'presence_penalty'
        - 'top_p'