Skip to content

Vision

Vision workflows use hosted image assets as model input. Pixapi image edit requests only accept image URLs in JSON fields.

Image editing

json
{
  "model": "gemini-3-pro-image-preview",
  "prompt": "Create a premium social ad using the same product and color palette.",
  "image": "https://cdn.example.com/product.png",
  "n": 1,
  "size": "1:1"
}
bash
curl https://api.pixapi.ai/v1/images/edits \
  -H "Authorization: Bearer $PIXAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3-pro-image-preview",
    "prompt": "Create a premium social ad using the same product and color palette.",
    "image": "https://cdn.example.com/product.png",
    "n": 1,
    "size": "1:1"
  }'

Input requirements

  • Send image inputs as hosted HTTPS URLs in the image field.
  • Do not send local files, multipart uploads, or base64 image data.
  • Prefer PNG, JPEG, or WebP assets.
  • Keep reference images relevant to the requested output.
  • Do not send private user files unless your application has permission to process them.

The older image-generation field names are available only in Legacy transition format.