GetRouterDocs
  • English

    • Tool Quick Start

    Most AI tools need three GetRouter values:

    Provider name: GetRouter
API key:       YOUR_API_KEY
Base URL:      https://getrouter.ai/v1
Model:         MODEL_ID

    Generic setup

    1. Open the tool's Providers, Models, or API settings.
    2. Select OpenAI-compatible, Custom OpenAI, or Custom Provider.
    3. Enter your GetRouter API key.
    4. Enter the Base URL expected by the tool.
    5. Paste an exact model ID from Models.
    6. Save the provider.
    7. Send a short test message.
    Warning

    The generic Base URL above is for OpenAI-compatible tools. Codex, Claude-compatible tools, and Gemini-compatible tools may require a different protocol or additional settings. Use the dedicated guide under Integrations.

    Determine what the tool expects

    Check the tool's provider settings or documentation:

    Tool fieldUsually enter
    OpenAI Base URL / API Hosthttps://getrouter.ai/v1
    Full Chat Completions endpointhttps://getrouter.ai/v1/chat/completions
    Responses endpointhttps://getrouter.ai/v1/responses
    Claude Messages endpointhttps://getrouter.ai/v1/messages
    Gemini API roothttps://getrouter.ai with the tool's Gemini path configuration

    Do not append an operation path if the tool already adds it.

    Verify the connection

    Use a small text request first. After it succeeds, test any advanced capability you need, such as streaming, reasoning, tool calling, vision, or media generation.

    If the tool can load models but cannot send a message, the selected model may not support the protocol used by that tool.

    Create one key per tool:

    Codex Laptop
OpenClaw Home
Chat Client
Automation

    This makes usage easier to identify and allows you to revoke one integration without affecting the others.

    Troubleshooting checklist

    • Remove spaces before or after the API key.
    • Check whether the tool automatically appends /v1.
    • Check whether it calls /chat/completions, /responses, or /messages.
    • Use the exact model ID from Models.
    • Restart the tool if it only reads environment variables at launch.
    • Review the dedicated Integration guides.

    Next steps