Model Overview

Review shared image-model fields, differences, and entry pages for each image model.

Current public image models all use POST /v1/generations. The main differences are enum ranges, defaults, and recommended use cases.

Read this family page first, then open the individual model page only after choosing a model.

Field	Type	Required	Description
`model`	`string`	Yes	Target image model ID
`prompt`	`string`	Yes	Image prompt
`reference_images`	`string[]`	No	Reference image URL list
`aspect_ratio`	`string`	No	Output aspect ratio
`resolution`	`string`	No	Output resolution
`n`	`integer`	No	Image count

Model	Region	Best For	`quality`	`aspect_ratio`	`resolution`	Defaults	Details
`gpt-image-2`	Global / CN	Best default general image model	Not supported	`auto` `1:1` `5:4` `9:16` `21:9` `16:9` `3:2` `4:3` `4:5` `3:4` `2:3`	`1K` `2K` `4K`	`aspect_ratio=auto` `resolution=1K` `n=1`	View
`gpt-image-2-official`	Global	Official GPT Image 2	`low` `medium` `high`	Model-schema driven	`1K` `2K` `4K`	`quality=medium` `resolution=1K` `n=1`	View
`gemini-3.1-flash-image-preview`	Global / CN	Fast image generation and editing	Not supported	Model-schema driven	`0.5K` `1K` `2K` `4K`	`resolution=1K` `n=1`	View
`gemini-3-pro-image-preview`	Global / CN	High-quality image generation and editing	Not supported	Model-schema driven	`1K` `2K` `4K`	`resolution=1K` `n=1`	View
`nano-banana-2`	Global / CN	Speed-first	Not supported	`auto` `21:9` `16:9` `3:2` `4:3` `1:1` `3:4` `2:3` `9:16`	`2K` `4K`	`aspect_ratio=auto` `resolution=2K` `n=1`	View
`nano-banana-pro`	Global / CN	Quality-first, polished creative	Not supported	`auto` `21:9` `16:9` `3:2` `4:3` `1:1` `3:4` `2:3` `9:16`	`2K` `4K`	`aspect_ratio=auto` `resolution=2K` `n=1`	View

How to Choose an Image Model

If you are unsure, start with gpt-image-2.
If speed matters more, try nano-banana-2 first.
If quality and editing matter more, try nano-banana-pro first.
Official GPT Image 2 exposes quality independently; do not send it to regular gpt-image-2.
Official quality tiers are low (speed/cost-first), medium (default balance), and high (quality-first, higher cost).
Use each Gemini image model's schema for resolution values; Flash supports 0.5K.

{
  "model": "gpt-image-2",
  "prompt": "A premium skincare poster on warm marble, soft daylight, luxury editorial style"
}