Inference API
Aurora Inference API lets you run large language models through an OpenAI-compatible API, with per-token billing and usage tracking. Connect coding agents, chat applications, and SDKs by configuring your Aurora base URL and API token. See the integration guides for OpenCode, Pi, Hermes Agent, OpenClaw, Chatbox AI, and more.
Base URL
OpenAI Compatibility
The API supports the following OpenAI-compatible operations:
| Endpoint | Description |
|---|---|
GET /v1/models | List default models, or tenant models when authenticated |
GET /v1/models/{model} | Get a default model, or a tenant model when authenticated |
POST /v1/chat/completions | Chat completions (streaming + non-streaming; authentication required) |
POST /v1/completions | Text completions (streaming + non-streaming; authentication required) |
Model discovery is public: both model endpoints work without authentication and return the models enabled by default. When a request carries an API token, they return the models available to your tenant instead. Chat and text completions require an API token. See Getting started for authentication and account prerequisites.
Chat and text completions support streaming via server-sent events (SSE). Tool calling, multimodal input, structured output, and sampling options depend on the selected model.
Compatibility is limited to the documented operations. OpenAI /v1/responses and
Anthropic /v1/messages are not supported.
Next Steps
- Getting Started — obtain an API token and make your first request
- Models and pricing — choose a model and understand usage costs
- API reference — request fields, responses, streaming, and errors
- OpenAPI specification — download the specification for clients and tools
- Integrations — connect coding agents, chat applications, and libraries
- OpenAI Python and JavaScript / TypeScript — use the official OpenAI SDKs