GPT Image 2

GPT Image 2 (openai/gpt-image-2) is OpenAI’s image generation model. It works with 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.

Parameters

Field	Description
model	openai/gpt-image-2 (or the short gpt-image-2)
prompt	What to draw, required ; up to 32,000 characters
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 from the model’s fixed table
quality	The model has no such knob: passing the parameter returns 400

GPT Image 2 sizes are not computed from the aspect ratio — they come from a table fixed per “resolution + aspect” pair. Common combinations:

Aspect	1k	2k	4k
1:1	1024x1024	2048x2048	2880x2880
16:9	1792x1024	2048x1152	3840x2160
9:16	1024x1792	1152x2048	2160x3840
4:3	1024x1024	2048x1536	3328x2496
21:9	1792x1024	2880x1232	4480x1920

Note the 1k column: several different aspects yield the same WxH. For this model it is safer to pass aspect_ratio + resolution and treat size as compatibility — on a collision, the reverse WxH → aspect mapping picks the first matching option.

Examples

Text-to-image

Python

import base64

img = client.images.generate(
    model="openai/gpt-image-2",
    prompt="Isometric illustration of a book cafe, warm palette, fine detail",
    extra_body={"aspect_ratio": "16:9", "resolution": "2k"},
)
with open("cafe.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": "openai/gpt-image-2",
    "prompt": "Isometric illustration of a book cafe, warm palette, fine detail",
    "aspect_ratio": "16:9",
    "resolution": "2k"
  }'

An explicit size instead of the pair

{
  "model": "openai/gpt-image-2",
  "prompt": "Podcast cover about space: minimalism, deep blue, stardust",
  "size": "2880x1232"
}

Editing from a reference

The model accepts reference images (references: true), so /v1/images/edits is available:

curl https://api.mixen.ai/v1/images/edits \
  -H "Authorization: Bearer $MIXEN_API_KEY" \
  -F model=openai/gpt-image-2 \
  -F image=@cat.png \
  -F prompt="Dress the cat as an astronaut, keep the background"

The answer always carries b64_json. Both endpoints answer synchronously — set your HTTP client timeout to at least 120 seconds.

Price

4.0108 ₽ ($0.045) per image — the flat catalog price. Current values — catalog.

Strengths and limits

• Twenty-three fixed sizes across three resolutions — from 1024x1024 up to 3840x2160 and 4480x1920.
• Ten frame formats, including the panoramic 21:9 for banners and covers.
• Editing from a reference via /v1/images/edits — the same key and syntax as generation.
• At 1k different aspects produce the same WxH — pass aspect_ratio + resolution, not size.
• No quality knob; a size outside the model’s table returns 400.
• Balance only — the model is not part of the free quota.

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