NEWThe new OpenAI image model in two modes: Flare for fast everyday work and Sunburst for precise final edits. Transparent backgrounds, crisp text on the image, up to 4K.
One model, two modes from OpenAI. The price per image is the same, so pick by the task, not by the budget.
Default mode
Higher quality than GPT Image 2 at 50% lower latency, according to OpenAI. The mode to start with.
In our test: 27-29 seconds per 2K image
Maximum precision
Extra precision for detailed work, at the cost of a longer generation. Holds the details while you edit.
In our test: 33-34 seconds on the same prompt
Tip
How we work ourselves: find the idea in Flare, then finish the final version in Sunburst.
We asked Sunburst to keep the face and turn a kitchen selfie into a 1970s movie frame. The face, stubble and smile stayed; the scene, clothes and light are new. The source photo is generated too: there are no real people in our examples.
Real results created by aiTurbo users
Stickers, icons and product cut-outs come without a background, ready to drop into a layout. Three settings: auto, transparent, opaque. A transparent frame takes longer: 75-96 seconds in our test.
Signs, labels, posters and packaging: letters stay crisp and spelled the way you wrote them, Cyrillic included.
Changes what you asked for and leaves the rest alone. Tip: ask to replace, not to remove. "Draw plain sneakers" works better than "remove the logo".
1K, 2K and 4K output. The square 1:1 frame goes to 4K too, 2880 by 2880 pixels.
From 21:9 banners to 9:16 stories, plus the rarer 27:16, 16:27, 9:8 and 8:9 in 1K.
Describe the scene in Russian and in detail. No translation step.
Select GPT Image 2.5 or any other from 35+ models
In any language - just describe what you want to see
Image or video ready in seconds. Download or use right away
We use the official API - generation quality is identical to the original
Enter a description - get results in seconds. No signup, no VPN.
Start generatingTry it and get 20 tokens free
