Image Generation (GPT-Image)

Text-to-Image

POST https://api.4allapi.com/v1/images/generations

curl https://api.4allapi.com/v1/images/generations \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AIPROXY_KEY" \
  -d '{
    "model": "gpt-image-2",
    "prompt": "Cyberpunk-style city night scene, neon lights, streets after rain",
    "n": 1,
    "size": "1024x1024"
  }'

Parameter	Description
`model`	Image model name, such as the `gpt-image-2` series
`prompt`	Description of the image; recommended to include subject, style, composition, and lighting
`n`	Number of images to generate
`size`	Resolution, such as `1024x1024`, `1536x1024` (landscape), `1024x1536` (portrait)
`quality`	Quality tier (supported models)

The response returns an image URL or base64 (b64_json); the URL is valid for a limited time, so please transfer it promptly.

Image-to-Image / Edit

POST https://api.4allapi.com/v1/images/edits

Upload a reference image as multipart/form-data:

curl https://api.4allapi.com/v1/images/edits \
  -H "Authorization: Bearer $AIPROXY_KEY" \
  -F model="gpt-image-2" \
  -F image="@input.png" \
  -F prompt="Replace the background with a seaside at dusk, keeping the person unchanged"

High-Resolution Variants

Some GPT-Image models provide 2K / 4K and landscape/portrait variants (subject to the models actually listed in the model marketplace, such as suffixes like -2k / -4k). Simply select the corresponding variant model name; no additional parameters are needed, and billing is based on variant pricing.