How Router works

Router gives applications one OpenAI-compatible connection to text models running across participating provider computers. Your application chooses the exact model edition. Router validates the request, finds free matching capacity, and relays the response.

The request path

1. Your application sends an authenticated request with an exact model ID.
2. Router validates the API key, account limits, request shape, and model limits.
3. Router selects one free provider that advertises the requested model revision.
4. The provider runs the prompt locally and returns ordered output.
5. Router returns a complete response or an ordered event stream to your application.

What Router decides

Router rotates work among eligible free providers for the model you selected. It tracks short-lived assignments, cancellations, leases, and ordered output so a request has one clear lifecycle.

What your application decides

Your application chooses the model, messages, output limit, sampling temperature, and whether to stream. Router does not automatically choose a model, silently substitute one, or keep requests in a waiting queue.

Failure stays explicit

If no matching provider is free, Router returns an error. If generation is interrupted, Router reports the interruption instead of combining partial text with another attempt. This gives your application control over retry timing and user-visible state.

Current trust boundary

Participating provider computers process request content and report token usage. Use non-sensitive prompts during the trusted-tester preview. Provider-reported usage is operational data, not verified billing data.