GPT Image 2 - GENGEN API

Image generation and reference-image editing with flexible output dimensions.

Model ID

gpt-image-2

Pass this exact public ID in the top-level model field. Do not substitute an upstream provider alias.

Supported modes

image_generation
image_edit

Inputs

Text
Image

Outputs

Image

Capabilities

Text-to-image
Image editing

Additional capabilities

Reference images
Transparent backgrounds (preview)

Technical details

Property	Value
Provider	`openai`
Modality	`image`
Processing	`Synchronous response`
Endpoint	`/api/gengen/v1/images/generations`
Example sizes (custom dimensions supported)	`1024x1024, 1536x1024, 1024x1536, 2048x2048, 2048x1152, 3840x2160, 2160x3840, auto`
Output delivery	`Public HTTPS URL only`
URL retention	`24 hours`
Quality	`low, medium, high, auto`
GENGEN defaults	`1024x1024, medium quality, 1 image, PNG, auto background`
Reference images	`Up to 16 HTTPS URLs; Base64 input is not supported`
Streaming	`Not supported by this GENGEN endpoint`
Watermark control	`Not supported`

Generated images are returned as public HTTPS URLs only. Each URL is retained for 24 hours; copy the image to storage you control before it expires.

Availability and final customer prices can change. Check the Dashboard model catalog before production use.

See the OpenAI image API reference for generation and editing examples, custom size constraints, request limits, and errors. GENGEN accepts only the model IDs listed in that reference, not dated OpenAI snapshots. Try GPT Image 2 in API Explorer. See the API reference overview for authentication, request conventions, and response schemas.