# LLM API

OpenAI-compatible endpoints that serve the models of a project. List models, create chat completions and responses, and answer typed questions.

The LLM API serves the models of a project over HTTP. Its endpoints are compatible with OpenAI's API.

The endpoints are at `https://data.spiceai.io`. Authorize each request with a [Project API key](https://spice.ai/docs/cloud/portal/apps/api-keys) in the `X-API-Key` header.

| Endpoint | Use | Reference |
| - | - | - |
| `GET /v1/models` | List the models of the project. | [Models](https://spice.ai/docs/cloud/api/openai-api/models) |
| `POST /v1/chat/completions` | Create a model response for a chat conversation. | [Chat Completions](https://spice.ai/docs/cloud/api/openai-api/chat-completions) |
| `POST /v1/responses` | Create a model response with OpenAI's Responses API format. | [Responses](https://spice.ai/docs/cloud/api/openai-api/responses) |
| `POST /v1/decisions` | Answer typed questions about an input, with probabilities. | [Decisions](https://spice.ai/docs/cloud/api/openai-api/decisions) |

Chat completions, responses, and decisions need a model that is configured and deployed in the project.