Crate chat_openrouter 

OpenRouter provider for chat-rs.

OpenRouter is a unified gateway in front of hundreds of models from many vendors; the model slug selects which one (e.g. anthropic/claude-sonnet-4, openai/gpt-4o, google/gemini-2.5-pro).

It exposes two OpenAI-compatible wires, and the builder picks one upstream and hands off to the matching wire crate — no wrapper type:

• Responses API (Beta) (default) — wraps chat_responses and builds a ResponsesClient. Stateless (no previous_response_id round-trip), so response-id reuse is disabled and the full conversation is sent each turn.
• Chat Completions — opt in with OpenRouterBuilder::with_completions; wraps chat_completions and builds a CompletionsClient.

Both wire clients already implement CompletionProvider (and StreamProvider under the stream feature), so streaming is free on either path. OpenRouter has no WebSocket endpoint — streaming is SSE over HTTP — but the builder stays generic over Transport.

use chat_openrouter::OpenRouterBuilder;

// OPENROUTER_API_KEY env var is read automatically.
// Default: Responses API.
let responses = OpenRouterBuilder::new()
    .with_model("anthropic/claude-sonnet-4")
    .build();

// Opt in to the Chat Completions API.
let completions = OpenRouterBuilder::new()
    .with_completions()
    .with_model("openai/gpt-4o")
    .build();

Structs

Completions

Wire type-state: target the Chat Completions API.

CompletionsClient

OpenRouterBuilder

ReqwestTransport

HTTP transport backed by reqwest.

Responses

Wire type-state: target the Responses API (default).

ResponsesClient

Generic client over the OpenAI Responses API wire format (POST {base_url}/responses). Provider wrappers like [chat_openai] hold one of these as their inner state.

WithModel

WithoutModel

Constants

DEFAULT_OPENROUTER_BASE_URL

Default OpenRouter base URL. The Responses API lives at {base_url}/responses, Chat Completions at {base_url}/chat/completions.