GPT Image 2 Image Editing

OmniRouters OpenAI image protocol reference for the gpt-image-2 model.

Endpoint

POST /v1/images/edits

Minimal request

curl https://omnirouters.com/v1/images/edits \
  -H "Authorization: Bearer $OMNIROUTERS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{\n  "model": "gpt-image-2",\n  "prompt": "Make the lighting warmer and preserve the composition.",\n  "image": "<binary image file>"\n}'

This operation edits an uploaded image. Use multipart form data when the gateway requires a binary upload.

Request body

Field	Type	Required	Description
model	string	必填	固定使用 gpt-image-2。
prompt	string	必填	Image prompt.
image	binary file	必填	待编辑的图片文件。
quality	string	否	图片质量：low、medium 或 high。
n	integer	否	生成数量。

The response may return Base64 data in data[].b64_json or a hosted URL in data[].url, depending on the model.

Response body

{
  "created": 1782875947,
  "data": [
    { "b64_json": "{image_base64_string}" }
  ]
}

Notes

Model-specific validation and output availability depend on the selected model and current OmniRouters routing configuration. For shared routing guidance, see Omni-Image API.