openai/gpt-image-2.6-sunburst

OfficialView on Replicate →

OpenAI's image model for detailed compositions and precise edits. Generate and edit images from text and image inputs, with a complexity setting that controls how much the model plans before it renders.

Capabilities

Reference Images

Cost

$0.500 / image

Input Parameters

Name	Type	Description	Default	Constraints
`prompt`*	string	A text description of the desired image	`—`	—
`aspect_ratio`	string	The size of the generated image. Use 'auto' to let the model pick. Sizes above 2560x1440 are experimental and may give more variable results.	`"1:1"`	1:13:22:34:33:416:99:16auto1024x10241536x10241024x15361536x11521152x15362048x20482048x11521152x20483840x21602160x3840
`background`	string	Set whether the background is transparent or opaque or choose automatically	`"auto"`	autotransparentopaque
`complexity`	string	How much the model plans before rendering. Higher complexity helps with dense text and complex layouts but can increase latency.	`"high"`	lowmediumhigh
`frame_count`	integer	Number of frames in a GIF. Ignored for other formats. For GIFs, named aspect ratios map to smaller frame sizes (1:1 is 512x512) because all frames share one image budget.	`9`	min: 2, max: 16
`frame_duration_ms`	integer	How long each GIF frame shows, in milliseconds, rounded to 10 ms. Ignored for other formats.	`200`	min: 20, max: 2000
`input_images`	array	A list of images to use as input for the generation	`—`	—
`moderation`	string	Content moderation level	`"auto"`	autolow
`number_of_images`	integer	Number of images to generate (1-10)	`1`	min: 1, max: 10
`openai_api_key`	string(password)	Your OpenAI API key (optional - uses proxy if not provided)	`—`	—
`output_compression`	integer	Compression level (0-100%)	`90`	min: 0, max: 100
`output_format`	string	Output format. 'gif' returns an animated GIF. 'svg' returns a vector image and works only for generation with one image.	`"webp"`	pngjpegwebpgifsvg
`quality`	string	The quality of the generated image. Higher quality takes longer and costs more.	`"auto"`	lowmediumhighxhighmaxauto
`svg_model`	string	SVG only. The model that traces the generated image into vector paths. gpt-6-astra and gpt-5.5 trace most carefully but are slower and cost more, and very detailed scenes can time out. Ignored for other formats.	`"gpt-5.6-terra"`	gpt-5.6-terragpt-6-astragpt-5.5gpt-5.4gpt-6-luna

promptrequiredstring

A text description of the desired image

aspect_ratiostring

The size of the generated image. Use 'auto' to let the model pick. Sizes above 2560x1440 are experimental and may give more variable results.

Default: "1:1"

1:13:22:34:33:416:99:16auto1024x10241536x10241024x15361536x11521152x15362048x20482048x11521152x20483840x21602160x3840

backgroundstring

Set whether the background is transparent or opaque or choose automatically

Default: "auto"

autotransparentopaque

complexitystring

How much the model plans before rendering. Higher complexity helps with dense text and complex layouts but can increase latency.

Default: "high"

lowmediumhigh

frame_countinteger

Number of frames in a GIF. Ignored for other formats. For GIFs, named aspect ratios map to smaller frame sizes (1:1 is 512x512) because all frames share one image budget.

Default: 9min: 2, max: 16

frame_duration_msinteger

How long each GIF frame shows, in milliseconds, rounded to 10 ms. Ignored for other formats.

Default: 200min: 20, max: 2000

input_imagesarray

A list of images to use as input for the generation

moderationstring

Content moderation level

Default: "auto"

autolow

number_of_imagesinteger

Number of images to generate (1-10)

Default: 1min: 1, max: 10

openai_api_keystring

Your OpenAI API key (optional - uses proxy if not provided)

output_compressioninteger

Compression level (0-100%)

Default: 90min: 0, max: 100

output_formatstring

Output format. 'gif' returns an animated GIF. 'svg' returns a vector image and works only for generation with one image.

Default: "webp"

pngjpegwebpgifsvg

qualitystring

The quality of the generated image. Higher quality takes longer and costs more.

Default: "auto"

lowmediumhighxhighmaxauto

svg_modelstring

SVG only. The model that traces the generated image into vector paths. gpt-6-astra and gpt-5.5 trace most carefully but are slower and cost more, and very detailed scenes can time out. Ignored for other formats.

Default: "gpt-5.6-terra"

gpt-5.6-terragpt-6-astragpt-5.5gpt-5.4gpt-6-luna

Version: 90b3d391b075Updated: 11/7/20267.4K runs