GPT Image 2

ImageOpenAI

OpenAI flagship image model — reasoning, near-perfect text rendering, flexible sizes up to 4K.

GPT Image. OpenAI image generation

Vendor: OpenAI
Family: GPT Image
Type: Image
Model ID
Tasks: fal.ai: Text to image, Image to image · Higgsfield: Text to image, Image to image · OpenAI: Text to image, Image to image

Prompt: fal.ai: Optional · Higgsfield: Optional · OpenAI: Optional
Image: fal.ai: Optional · Higgsfield: Optional · OpenAI: Optional
Max images: fal.ai: 1 · Higgsfield: 5 · OpenAI: 1
Reference images: fal.ai: Up to 1 reference image. · Higgsfield: Up to 5 reference images. · OpenAI: Up to 1 reference image.

Aspect ratio: fal.ai: Supported. The catalog does not list ratios. · OpenAI: Supported. The catalog does not list ratios.
Image count: fal.ai: Up to 4. · Higgsfield: Up to 4. · OpenAI: Up to 4.
Resolution: Higgsfield: 1k, 2k, 4k
Aspect ratios: Higgsfield: 1:1, 4:3, 3:4, 16:9, 9:16, 3:2, 2:3
Quality: Higgsfield: low, medium, high · OpenAI: Can be set. The catalog does not list choices.

The catalog has no structured price for this model.

Where to run it

Tasks and limits are per catalog row. A provider may offer only some of them.

Only sentences in the catalog that state an amount and a unit.

The catalog has no structured price for this model.

A still with rendered words
A small shop sign reading NORTH LIGHT, painted wood, overcast day
A change to an existing image
Keep the subject and shift the light to late afternoon

Every catalog row lists text to image and image to image, and allows up to 4 images in one request.