# POST /v1/chat/completions

> Request and response reference for Chat Completions.

```http
POST https://api.kurrens.ai/v1/chat/completions
Authorization: Bearer <KURRENS_API_KEY>
Content-Type: application/json
```

## Request body

| Field | Type | Required | Description |
|---|---|:--:|---|
| `model` | string | ✓ | Model id, e.g. `deepseek-ai/DeepSeek-V4.1-Flash` |
| `messages` | array | ✓ | Conversation so far: `system`/`developer`, `user`, `assistant`, `tool` messages |
| `max_tokens` | integer | | Max tokens to generate (includes reasoning tokens) |
| `temperature` | number | | Sampling temperature |
| `top_p` | number | | Nucleus sampling |
| `stop` | string \| string[] | | Up to 4 stop sequences |
| `stream` | boolean | | Stream server-sent events |
| `stream_options` | object | | `{ "include_usage": true }` |
| `tools` | array | | Function definitions |
| `tool_choice` | string \| object | | `auto`, `none`, `required`, or a named function |
| `response_format` | object | | `json_object` or `json_schema` |
| `reasoning_effort` | string | | `low`, `medium`, `high` on reasoning models |
| `seed` | integer | | Best-effort determinism |
| `service_tier` | string | | `priority` or `flex` where offered |

Per-model support and ranges are declared in [models.json](/models.json).

## Response

```json
{
  "id": "chatcmpl-…",
  "object": "chat.completion",
  "created": 1790640000,
  "model": "deepseek-ai/DeepSeek-V4.1-Flash",
  "choices": [
    {
      "index": 0,
      "message": { "role": "assistant", "content": "…" },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 24,
    "completion_tokens": 31,
    "total_tokens": 55,
    "prompt_tokens_details": { "cached_tokens": 0 },
    "completion_tokens_details": { "reasoning_tokens": 0 }
  }
}
```

`finish_reason` is one of `stop`, `length`, `tool_calls`, or `content_filter`.

Errors are described in [Errors](/docs/getting-started/errors).