Nano Banana Pro

Nano Banana Pro (google/gemini-3-pro-image-preview) is Google’s flagship model for photorealistic images up to 4K. Two endpoints: POST /v1/images/generations for text-to-image and POST /v1/images/edits for working from a reference image. The model is paid — generations are charged from your balance; there is no free quota for it.

Parameters

Field	Description
model	google/gemini-3-pro-image-preview
prompt	What to draw, required
n	How many images, 1 to 4; each one is billed separately
aspect_ratio	1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9; defaults to 1:1
resolution	1k, 2k or 4k; defaults to 1k
size	Explicit WxH — an alternative to the aspect_ratio + resolution pair
quality	The model has no such knob: passing the parameter returns 400

The model thinks in “aspect ratio + resolution” terms, and WxH is derived from them. Mappings for common combinations:

Aspect	1k	2k	4k
1:1	1000x1000	2000x2000	2828x2828
16:9	1332x750	2666x1500	3770x2120
9:16	750x1332	1500x2666	2120x3770

The full list of 30 allowed size values is the sizes field in the catalog; a value outside the list returns 400.

Examples

Text-to-image

Python

import base64

img = client.images.generate(
    model="google/gemini-3-pro-image-preview",
    prompt="Portrait of a jeweler at work, soft side light, shallow depth of field",
    extra_body={"aspect_ratio": "4:5", "resolution": "2k"},
)
with open("jeweler.png", "wb") as f:
    f.write(base64.b64decode(img.data[0].b64_json))

cURL

curl https://api.mixen.ai/v1/images/generations \
  -H "Authorization: Bearer $MIXEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "google/gemini-3-pro-image-preview",
    "prompt": "Portrait of a jeweler at work, soft side light, shallow depth of field",
    "aspect_ratio": "4:5",
    "resolution": "2k"
  }'

A 4K image for a cover

{
  "model": "google/gemini-3-pro-image-preview",
  "prompt": "Wide shot: fog over a coniferous valley, first sunrays, photorealism",
  "aspect_ratio": "21:9",
  "resolution": "4k"
}

Editing from a reference

The model accepts reference images (references: true), so /v1/images/edits with multipart/form-data is available:

curl https://api.mixen.ai/v1/images/edits \
  -H "Authorization: Bearer $MIXEN_API_KEY" \
  -F model=google/gemini-3-pro-image-preview \
  -F image=@portrait.png \
  -F prompt="Replace the background with an evening city, keep the face and the light"

The answer always carries b64_json — files are not hosted. Both endpoints answer synchronously; set your HTTP client timeout to at least 120 seconds.

Price

8.0216 ₽ ($0.09) per image — the flat catalog price. Current values — catalog.

Strengths and limits

• Photorealistic images are the model’s core profile, with 4K as the top step.
• Ten frame formats: from square 1:1 to panoramic 21:9, across three resolutions.
• Editing from a reference: style transfer and targeted fixes via /v1/images/edits.
• Balance only — the model is not part of the free quota.
• No quality knob; a size outside the model’s table returns 400.
• When transferring style, the model inherits the reference’s colour palette — describe a different palette explicitly in the prompt.

All models — in the catalog. The general image workflow — images guide.