API overview

primerouter exposes a REST API fully compatible with OpenAI, plus native support for Anthropic Messages and Google Gemini.

Base URL

https://primerouter.ai

OpenAI-style endpoints are under /v1/.... Claude-style endpoints are listed in API Reference.

Authentication

Authorization: Bearer sk-your-key

Or Anthropic-style:

x-api-key: sk-your-key

Get a key: log in → Tokens → create.

Main endpoints

Endpoint	Use
POST /v1/chat/completions	Multi-turn chat (OpenAI-compatible, most common)
POST /v1/responses	OpenAI Responses API
POST /v1/messages	Anthropic Messages (native Claude)
POST /v1/embeddings	Vector embeddings
POST /v1/images/generations	Text to image
POST /v1/images/edits	Image editing, multiple reference images
POST /v1/audio/speech	TTS
POST /v1/audio/transcriptions	Whisper STT
GET /v1/models	List models
POST /v1/rerank	Document reranking

Full schema and examples: API Reference (interactive viewer).

Sub-topics

• Authentication — key types, expiry, IP allowlist
• Streaming — SSE protocol details
• Error codes — standard error structure and handling
• Images — multiple reference images, size/quality limits, per-image billing
• Image Generation Skill — website download, cross-agent installation, and usage
• API Reference — interactive endpoint browser (try it live)

SDKs and examples

primerouter does not ship a dedicated SDK — the goal is for you to keep using your existing SDK. See Migrate from OpenAI SDK.

Rate limits

Each API key is bound by:

• User-level RPM / TPM (depends on your group)
• Model-level RPM (some models have upstream limits)

Hitting the limit returns 429 Too Many Requests with a Retry-After header. See Error codes.