Flare for iteration
Fast drafts and batch jobs — as quick as six seconds — with layout quality tuned for professional visuals.
OpenAI
Generate and edit with GPT Image 2.6
OpenAI’s fastest image generation yet — Flare for speed, Sunburst for refined quality and edit consistency, Economy for budget 1K output. Up to 16 reference images.
GPT Image 2.6 Studio
flare · sunburst · economy
Describe any image — structured commercial graphics, product imagery, posters — and GPT Image 2.6 renders it with strong layout quality.
Click or drop images here
1–16 images (JPG/PNG/WEBP, max 30MB each). Refer to them by order in the prompt, e.g. “put the product from image 1 into the scene of image 2”.Up to 20,000 characters. Sunburst keeps characters and layouts consistent across sequential revisions.
Estimated cost 10 gems
Your image will appear here
What is GPT Image 2.6?
OpenAI’s Images 2.6 family — faster generation, stronger layouts in Flare and more consistent editing in Sunburst.
50% fasterThan the previous version
16 refsReference-guided editing
3 tiersFlare / Sunburst / Economy
Up to 4K15 aspect ratios
Why creators use GPT Image 2.6
Pick the temperament: speed, quality or budget — the request format never changes.
Sunburst for refinement
Higher detail and even greater consistency across sequential edits — characters and layouts hold steady.
Economy for scale
A budget tier at 1K with the core aspect ratios — same request format, lowest friction.
Reference variations
Build new versions of a product, character or scene from up to 16 of your own source images.
Transparent cutouts
Product cutouts, stickers and isolated assets at 1K with background set to transparent.
One-line migration
All tiers share one request schema and one price per resolution — switch by changing the model ID.
How to create an image
Pick a tier and mode, configure the output and generate.
1. Choose tier & mode
Flare, Sunburst or Economy; Text to Image or Image to Image with references.
2. Write the prompt
Describe the visual — or the edit, naming the reference images by order.
3. Configure the output
Aspect ratio, resolution (Economy locks to 1K) and background.
4. Generate and download
The studio estimates your cost live. Preview the result inline and download the image.
GPT Image 2.6 FAQ
Answers to common questions about tiers, inputs and pricing.
What is GPT Image 2.6?
GPT Image 2.6 is OpenAI’s latest image model family, released November 4, 2026. It generates images up to 50% faster than before and comes in Flare (the fast version), Sunburst (refined quality and edit consistency) and Economy (1K-only budget tier).
What is the difference between Flare, Sunburst and Economy?
Flare is optimized for speed — as fast as six seconds — with layout quality tuned for posters and product imagery. Sunburst is built for higher detail and stronger consistency across multi-round edits. Economy is the budget tier: 1K resolution only, with a reduced set of aspect ratios.
What is the difference between Text to Image and Image to Image?
Text to Image generates a new picture from your prompt. Image to Image takes 1 to 16 reference images (input_urls) and follows your instruction — refer to them by order, for example “put the product from image 1 into the background of image 2”.
What resolutions and aspect ratios are supported?
Flare and Sunburst: 1K, 2K or 4K with 15 aspect ratios plus auto. Economy: 1K only, with auto, 1:1, 16:9, 9:16, 3:2, 2:3, 21:9 and 4:3. Background transparency is available at 1K.
How does billing work?
A flat per-image rate by resolution: 10 gems for 1K, 20 for 2K and 40 for 4K — the same rate for all three tiers. Reference images are free.
Can I get transparent backgrounds?
Yes. Set background to transparent at 1K output for product cutouts, stickers and isolated design assets.
How many images do I get per request?
One image per request — use the seed or iterate with references for variations. Prompts can be up to 20,000 characters.
Ready to create your next visual?
Start with a prompt or a set of references and generate images with GPT Image 2.6.
Start creating with GPT Image 2.6