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

# OpenAI chat format

## OpenAI chat format (Chat Completions + Gemini image)

Standard OpenAI Chat Completions, with chat-style prompts driving Gemini image generation.

***

### Basics

* **Endpoint**: `POST /v1/chat/completions`
* **Auth**: `Bearer Token`
* **Content-Type**: `application/json`

***

### Request body

| Field                      | Type      | Required | Description                                  |
| :------------------------- | :-------- | :------- | :------------------------------------------- |
| **model**                  | `string`  | **Yes**  | e.g. `gemini-3-pro-image-preview`.           |
| **messages**               | `array`   | **Yes**  | Chat messages.                               |
| └─ **role**                | `string`  | **Yes**  | `user`, `assistant`, `system`.               |
| └─ **content**             | `string`  | **Yes**  | Prompt, e.g. “draw a cat”.                   |
| **stream**                 | `boolean` | **Yes**  | Streaming; use `false` for image generation. |
| **extra\_body**            | `object`  | No       | Vendor extensions.                           |
| └─ **google**              | `object`  | No       | Google-specific block.                       |
| └─ └─ **image\_config**    | `object`  | No       | Image settings.                              |
| └─ └─ └─ **aspect\_ratio** | `string`  | No       | e.g. `16:9`, `1:1`.                          |
| └─ └─ └─ **image\_size**   | `string`  | No       | e.g. `2K`, `4K`.                             |

***

### Response

**200 OK**

| Field                 | Type     | Description                                            |
| :-------------------- | :------- | :----------------------------------------------------- |
| **id**                | `string` | Request id.                                            |
| **object**            | `string` | `chat.completion`.                                     |
| **choices**           | `array`  | Results.                                               |
| └─ **message**        | `object` | `role` + `content` (image often as Markdown + base64). |
| └─ **finish\_reason** | `string` | e.g. `stop`.                                           |
| **usage**             | `object` | Token usage.                                           |

***

### Request example

```json theme={null}
{
  "model": "gemini-3-pro-image-preview",
  "stream": false,
  "messages": [
    {
      "role": "user",
      "content": "draw a cat"
    }
  ],
  "extra_body": {
    "google": {
      "image_config": {
        "aspect_ratio": "16:9",
        "image_size": "2K"
      }
    }
  }
}
```

### Response example

```json theme={null}
{
  "id": "chatcmpl-20251211...",
  "model": "gemini-3-pro-image-preview",
  "object": "chat.completion",
  "created": 1765440499,
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "![image](data:image/jpeg;base64,/9j/4AAQSk...)"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 4,
    "completion_tokens": 1359,
    "total_tokens": 1363
  }
}
```

***

### Tips

1. **Extracting images**: `content` often embeds `![image](data:image/jpeg;base64,...)`—parse the data URI for display.
2. **extra\_body**: Standard OpenAI SDK extension slot for vendor options without breaking the protocol.
3. **Mock**: For local tests, `http://127.0.0.1:4523/mock/7707339`.