GPT-Image 2
OpenAI's Latest and Most Realistic Image Model
OpenAI's newest image model delivers photoreal portraits with extraordinary skin, hair, and lighting detail. Choose quality tier (low / medium / high) and resolution (1K / 2K / 4K), use your characters or reference images, and render up to 4 images per generation.
Try GPT-Image 2 NowExample Generations
Technical Capabilities
Resolution
1K / 2K / 4K
Quality Settings
Low / Medium / High
Reference Images
Up to 10
Character Reference
Yes
Max Images Per Generation
4
Provider
OpenAI
LoRA Support
No
Key Features
OpenAI's newest and most realistic image model
Three quality tiers from fast drafts to high-detail finals
Native 1K, 2K, and 4K output resolutions
Use your saved characters as reference images
Up to 10 reference images per generation
Excellent photoreal skin, hair, and lighting
Renders up to 4 images per request
How It Works
Describe Your Scene
Write a detailed prompt — GPT-Image 2 follows complex, multi-part instructions.
Pick Quality and Resolution
Choose Low, Medium, or High quality and pick a 1K, 2K, or 4K resolution tier.
Add Characters or References (Optional)
Attach a saved character or up to 10 reference images to lock identity and style.
Generate
Get up to 4 highly photoreal images in one request.
