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

# GPT-Image-2 Official Channel Image Editing

> Edit images, blend references, or apply inpainting with gpt-image-2 using the same endpoint as generation

* Uses the same `/v1/images/generations` endpoint as text-to-image generation
* Editing mode is triggered automatically when `image_urls` is included in the request
* Supports multi-reference blending (up to 16 images) and mask-based inpainting
* Reference images must be publicly accessible PNG / JPG, max 50MB each

<Note>
  Image editing is fully compatible with the generation API: simply add `image_urls` (and optionally `mask_url`) to your generation request — no endpoint switch required.
</Note>

## Authorizations

<ParamField header="Authorization" type="string" required>
  All endpoints require Bearer Token authentication

  Get your API Key from the [API Key Management Page](https://toapis.com/console/token)

  ```
  Authorization: Bearer YOUR_API_KEY
  ```
</ParamField>

## Body

<ParamField body="model" type="string" default="gpt-image-2" required>
  Use `gpt-image-2`
</ParamField>

<ParamField body="prompt" type="string" required>
  Describe the desired edit

  Example: `"Replace the background with a starry sky, keep the subject unchanged"`
</ParamField>

<ParamField body="image_urls" type="string[]" required>
  Reference image URL array (the field that triggers editing mode)

  * Maximum **16 images**
  * Must be publicly accessible stable URLs (PNG / JPG, max 50MB each)
  * Use the [Upload 画像API](../../uploads/images) to get URLs for local files
</ParamField>

<ParamField body="mask_url" type="string">
  Mask image URL for inpainting

  * Must be PNG format with an Alpha channel
  * Transparent areas (alpha = 0) will be repainted; opaque areas are preserved
  * Must match the dimensions of the first reference image
</ParamField>

<ParamField body="size" type="string" default="1:1">
  Output aspect ratio

  Supports 13 ratios or `auto`:

  `1:1` · `3:2` · `2:3` · `4:3` · `3:4` · `5:4` · `4:5` · `16:9` · `9:16` · `2:1` · `1:2` · `21:9` · `9:21`
</ParamField>

<ParamField body="resolution" type="string" default="1k">
  Resolution tier: `1k` / `2k` / `4k`

  1K and 2K support all 13 ratios. 4K only supports 6 ratios: `16:9` / `9:16` / `2:1` / `1:2` / `21:9` / `9:21`
</ParamField>

<ParamField body="quality" type="string" default="high">
  Image quality: `low` / `medium` / `high`
</ParamField>

<ParamField body="n" type="integer" default={1}>
  Number of images to generate, range `1 ~ 10`
</ParamField>

<ParamField body="output_format" type="string" default="png">
  Output format: `png` / `jpeg`

  <Note>Azure OpenAI does not support `webp` format.</Note>
</ParamField>

<ParamField body="output_compression" type="integer" default={100}>
  JPEG compression level, range `0–100` (only applies when `output_format` is `"jpeg"`)
</ParamField>

## レスポンス

<ResponseField name="id" type="string">
  Unique task identifier for status queries
</ResponseField>

<ResponseField name="object" type="string">
  Always `generation.task`
</ResponseField>

<ResponseField name="status" type="string">
  Task status: `queued` / `in_progress` / `completed` / `failed`
</ResponseField>

<ResponseField name="progress" type="integer">
  Task progress percentage (0-100)
</ResponseField>

<ResponseField name="created_at" type="integer">
  Task creation timestamp (Unix timestamp)
</ResponseField>

<RequestExample>
  ```bash Image-to-Image (Multi-reference Blend) theme={null}
  curl --request POST \
    --url https://toapis.com/v1/images/generations \
    --header 'Authorization: Bearer <token>' \
    --header 'Content-Type: application/json' \
    --data '{
      "model": "gpt-image-2",
      "prompt": "Blend these two references into a cyberpunk illustrated poster",
      "image_urls": [
        "https://example.com/input-a.png",
        "https://example.com/input-b.png"
      ],
      "size": "1:1",
      "quality": "high",
      "n": 1
    }'
  ```

  ```bash Inpainting (Mask) theme={null}
  curl --request POST \
    --url https://toapis.com/v1/images/generations \
    --header 'Authorization: Bearer <token>' \
    --header 'Content-Type: application/json' \
    --data '{
      "model": "gpt-image-2",
      "prompt": "Replace the background with a desert sunset, keep the person unchanged",
      "image_urls": ["https://example.com/photo.png"],
      "mask_url": "https://example.com/mask.png",
      "size": "1:1",
      "quality": "medium"
    }'
  ```

  ```bash Style Transfer (High-res Output) theme={null}
  curl --request POST \
    --url https://toapis.com/v1/images/generations \
    --header 'Authorization: Bearer <token>' \
    --header 'Content-Type: application/json' \
    --data '{
      "model": "gpt-image-2",
      "prompt": "Convert this photo to watercolor style, preserve the scene composition",
      "image_urls": ["https://example.com/landscape.jpg"],
      "size": "16:9",
      "resolution": "2k",
      "quality": "high",
      "output_format": "jpeg",
      "output_compression": 85
    }'
  ```

  ```javascript JavaScript theme={null}
  const response = await fetch('https://toapis.com/v1/images/generations', {
    method: 'POST',
    headers: {
      'Authorization': 'Bearer your-ToAPIs-key',
      'Content-Type': 'application/json',
    },
    body: JSON.stringify({
      model: 'gpt-image-2',
      prompt: 'Blend these two references into a cyberpunk illustrated poster',
      image_urls: [
        'https://example.com/input-a.png',
        'https://example.com/input-b.png',
      ],
      size: '1:1',
      quality: 'high',
      n: 1,
    }),
  });

  const task = await response.json();
  console.log(task.id, task.status);
  ```
</RequestExample>

<ResponseExample>
  ```json 200 theme={null}
  {
    "id": "tsk_img_01KPTXXXXXXXXXXXXXXX",
    "object": "generation.task",
    "model": "gpt-image-2",
    "status": "queued",
    "progress": 0,
    "created_at": 1703884800,
    "metadata": {}
  }
  ```
</ResponseExample>