For the complete documentation index, see llms.txt. Markdown versions of all docs pages are available by appending .md to any docs URL.

LLM API endpoint types and the route types that serve them.

Serve clients that send a particular LLM API format.

Clients reach a model through a specific API endpoint, such as the OpenAI Chat Completions endpoint or the native Gemini models/{model}:generateContent endpoint. Each endpoint has a matching route type that you set in the policies.ai.routes field of an AgentgatewayBackend resource. Requests default to the Completions route type, so an endpoint in another format must be mapped explicitly.

This section covers the API types that take configuration of their own. For the full list of route types, and to map several endpoints on the same AgentgatewayBackend, see Multiple endpoints.