> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fastrun.si/llms.txt
> Use this file to discover all available pages before exploring further.

# API reference

> Base URL, authentication and the endpoints of the fastrun API.

The fastrun API follows OpenAI's API. Requests and responses have the same shape, so OpenAI SDKs and tools work with
only the base URL, key and model changed.

## Base URL

```text theme={null}
https://api.fastrun.si/v1
```

HTTPS only, TLS 1.2 or later.

## Authentication

Send your API key as a bearer token. See [Authentication](https://docs.fastrun.si/authentication.md).

```http theme={null}
Authorization: Bearer fr_...
```

## Endpoints

| Method | Path | |
| - | - | - |
| `POST` | `/chat/completions` | [Create a chat completion](https://docs.fastrun.si/api-reference/chat/create-a-chat-completion.md) |
| `POST` | `/completions` | [Create a completion](https://docs.fastrun.si/api-reference/completions/create-a-completion.md) |
| `GET` | `/models` | [List models](https://docs.fastrun.si/api-reference/models/list-models.md) |

Other OpenAI endpoints, such as embeddings, images and the Responses API, are not supported.

## Requests

* Send JSON with `Content-Type: application/json`. Bodies can be up to 64 MiB.
* A streamed request can run for up to 60 minutes; one without streaming must be complete within 10 minutes. See
  [Timeouts](https://docs.fastrun.si/guides/errors.md#timeouts).
* Errors use OpenAI's error format; see [Errors and retries](https://docs.fastrun.si/guides/errors.md).

## Differences from OpenAI

* Model IDs are fastrun's; see [Models](https://docs.fastrun.si/models.md).
* Input is text only.
* Reasoning text comes in `message.reasoning` (and `delta.reasoning` when streaming). See [Reasoning](https://docs.fastrun.si/guides/reasoning.md).
* Extra sampling parameters are accepted: `top_k`, `min_p`, `repetition_penalty` and `chat_template_kwargs`.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.