Text-to-Image/Edit API Reference

API reference for the Nano Banana series (Gemini ecosystem image models). For series introduction and prompt tips, see Nano Banana Image Generation.

Text-to-Image

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

curl https://api.4allapi.com/v1/images/generations \
  -H "Authorization: Bearer $AIPROXY_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3.1-flash-image-portrait",
    "prompt": "Portrait with magazine-cover quality, natural light, shallow depth of field, with \"OMNI\" written at the bottom of the image"
  }'

Use model suffixes for aspect ratio and resolution: -square / -portrait / -landscape / -three-four / -four-three, and add -2k for high definition;
Gemini models perform well at rendering text in images; put the text content in quotes in the prompt.

Image Editing (Image-to-Image)

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

curl https://api.4allapi.com/v1/images/edits \
  -H "Authorization: Bearer $AIPROXY_KEY" \
  -F model="gemini-3.1-flash-image" \
  -F image="@ref.png" \
  -F prompt="Change the same person to a full-body winter snowy scene while keeping facial features consistent"

Pass multiple image fields in order for multiple reference images, and describe each image’s role in the prompt (subject image/style image);
For character/product consistency scenarios, prefer the editing API instead of regenerating with text-to-image.

Model Tiers

Tier	Representative Model	Features
Pro	`gemini-3.0-pro-image` series	Highest quality, most stable details and text
Standard	`gemini-3.1-flash-image` series	Balanced quality/speed, main tier
Preview	`gemini-3-pro-image`	Early access to new capabilities

For currently available models and pricing, refer to Model Marketplace; response structure and upload/storage notes are the same as in Text-to-Image API Reference.