Strix API overview

Strix exposes OpenAI-compatible Chat Completions and Responses plus Anthropic-compatible Messages at https://api.strixgate.dev. Send a canonical strix-* model ID and Strix selects the configured lane; strix-auto classifies each request before selecting a lane. Virtual ids are never forwarded upstream — Strix maps them to a real OpenRouter model (or Venice, when you stored a Venice key).

You must bring an OpenRouter key. Strix does not run customer inference on the platform OpenRouter credential.

Start here

1. Connect a Strix API key and store your OpenRouter key on Billing.
2. Point any supported OpenAI-compatible or Anthropic client at Strix. Setup cards cover Cursor, Claude Code, Codex, OpenCode, Cline, Roo Code, Continue, and Aider. See quickstart and pricing.
3. Follow the quickstart for your first completion.
4. Choose an intent in the model and lane guide, or let strix-auto route. Add strix-quick as Fast mode for small, focused tasks.
5. Inspect the x-strix-* response headers in errors and routing evidence. Optional memory remembers facts across sessions for any connected client.
6. Compare Strix to OpenRouter, Cursor, or alternatives if you are choosing a stack.

Core endpoints

• GET /v1/models — authenticated OpenAI-style model list.
• POST /v1/chat/completions — OpenAI-compatible chat, including SSE streaming.
• POST /v1/responses — stateless OpenAI Responses compatibility for Codex, including SSE streaming and function calls.
• POST /v1/messages — Anthropic message format, including translated SSE streaming.
• GET /v1/me — identity associated with the current key.
• GET /v1/usage — usage report for the current account.

Use the interactive playground to call these endpoints directly from your browser. The playground keeps the key in session storage and does not insert it into generated examples.