Endpoint conventions
Starmodel exposes a single OpenAI-compatible gateway. Every request hits one base URL; the path selects the capability and the model field selects the model.
Base URL
https://api.starmodel.ai/v1This is your gateway domain. Append the endpoint path below, and authenticate every request with Authorization: Bearer $API_KEY.
Endpoints
| Capability | Method | Path |
|---|---|---|
| Chat completions | POST | /v1/chat/completions |
| Anthropic messages | POST | /v1/messages |
| Responses API | POST | /v1/responses |
| Generate | POST | /v1/images/generations |
| Edit | POST | /v1/images/edits |
| Generate video | POST | /v1/videos/generations |
| Speech | POST | /v1/audio/speech |
| List models | GET | /v1/models |
Synchronous vs asynchronous
Chat, messages and responses reply within a single HTTP request. Set stream: true to receive Server-Sent Events (text/event-stream), terminated by data: [DONE].
Image, video and speech generation are long-running upstream, but the gateway polls the task for you and returns a unified synchronous JSON result — there is no task id to poll. Raise your client timeout for video. Rate-limited requests return HTTP 429 with a Retry-After header.
Migrating from a direct provider
Endpoints are OpenAI-compatible, so in most cases you only change the base_url:
- client = OpenAI(api_key=OPENAI_KEY)+ client = OpenAI(+ api_key=SEEAI_KEY,+ base_url="https://api.starmodel.ai/v1",+ )Using the Anthropic SDK? Point its base_url at the same gateway and call /v1/messages.