gpt-image-2.5

GPT-Image-2.5 uses the unified asynchronous task API. Submit a task to /api/generate/submit, save the returned platform task_id, then poll /api/generate/status/:task_id for the result.

• gpt-image-2.5-flare prioritizes speed for everyday generation, batches, and rapid prototyping.
• gpt-image-2.5-sunburst prioritizes editing precision for product imagery and production-ready creative work.
• Put every request parameter at the top level of the JSON body. Reference images must be public HTTP(S) URLs.
• n accepts 1–4. When multiple images are generated, data.files contains every result.

Both models support localized editing with mask_url, accepting a public HTTP(S) URL or PNG Data URL. The mask must have an alpha channel and match the first reference image; fully transparent areas mark edits. image and images together accept up to 16 references. Masks have no effect without a reference image.

curl -X POST "https://api.apimp.ai/api/generate/submit" \  -H "Content-Type: application/json" \  -d '{    "model": "gpt-image-2.5-sunburst",    "prompt": "仅将蒙版透明区域替换为金色丝带",    "images": [      "https://example.com/product.png"    ],    "mask_url": "https://example.com/mask.png",    "quality": "xhigh"  }'

{
  "code": 200,
  "data": {
    "task_id": "task_01KXXXXXXXXXXXXXXX",
    "status": "not_started",
    "created_time": "2026-09-09T10:30:00",
    "error_message": null
  }
}

{
  "code": 400,
  "error": {
    "message": "string",
    "type": "invalid_request_error"
  }
}

{
  "code": 400,
  "error": {
    "message": "string",
    "type": "invalid_request_error"
  }
}

{
  "code": 400,
  "error": {
    "message": "string",
    "type": "invalid_request_error"
  }
}

{
  "code": 400,
  "error": {
    "message": "string",
    "type": "invalid_request_error"
  }
}

{
  "code": 400,
  "error": {
    "message": "string",
    "type": "invalid_request_error"
  }
}