API overview

The primary rout.my API is OpenAI-compatible. Most SDKs only need a custom base URL and a rout.my API key.

https://api.rout.my/v1

Authentication

Use a bearer token for OpenAI-compatible endpoints:

Authorization: Bearer sk_your_key_here

Some compatibility routes also accept x-api-key or ?key=... because specific clients expect those styles. Bearer auth is the default for new integrations.

Endpoint families

Endpoint	Method	Format
/v1/models	GET	OpenAI-style model list.
/v1/chat/completions	POST	OpenAI Chat Completions.
/v1/responses	POST	OpenAI Responses-style request and stream shape.
/v1/embeddings	POST	OpenAI Embeddings.
/v1/images/generations	POST	OpenAI Images-style response.
/v1/video/generations	POST	OpenAI-ish video generation response.
/v1/messages	POST	Anthropic Messages-compatible route.
/v1/messages/count_tokens	POST	Anthropic Messages token-counting route.
/compatible/v1beta/models/{model}:generateContent	POST	Gemini-compatible route.
/embed/image:{model}	GET	Raw image bytes for embedding in pages or chats.

Minimal request

curl https://api.rout.my/v1/chat/completions \
  -H "Authorization: Bearer $ROUTMY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-5.4",
    "messages": [
      { "role": "user", "content": "Hello." }
    ]
  }'

CORS

The public API endpoints send permissive CORS headers for browser clients. Keep API keys out of public frontend code unless you intentionally want users to bring their own key.

Response shape

Successful responses are normalized to the endpoint family you called. Errors use a consistent JSON object:

{
  "error": {
    "message": "Model is required",
    "type": "invalid_request_error",
    "code": "Bad Request"
  }
}

See Errors for status codes and retry behavior.

Video compatibility

/v1/video/generations accepts a unified video request shape for text-to-video, image-to-video, reference-to-video, motion-control, and video editing. Use common fields such as duration, resolution, aspect_ratio, audio, quality, negative_prompt, input_image, input_video, last_frame_image, and reference_urls; rout.my maps them to provider-specific Vercel AI Gateway options when the selected model supports them.

See Video for the full field list and examples.