Flare for speed, Sunburst for quality. Up to 16 reference images, new xhigh / max quality tiers, and multi-turn edits that change only what you ask for.
New users get free credits — no credit card required.
GPT Image 2.5 is the direct successor to GPT Image 2. These are the changes that matter.
Flare targets speed with quality roughly on par with GPT Image 2. Sunburst targets higher fidelity and editing precision. Same credit price for both, so you pick by latency vs. polish.
The headline gain over GPT Image 2. Actual time depends on quality tier and size — treat it as a direction, not a guarantee.
Edit mode accepts up to 16 reference images per request, with the mask applied to the first one. That is what lets series work — campaign variants, storyboards, product sets — hold one identity across the batch.
On top of low / medium / high, 2.5 adds xhigh and max. Custom WIDTHxHEIGHT sizes are supported (multiples of 16, up to 3:1 aspect, up to 3840 px on a side), plus transparent PNG / WebP output.
The value of GPT Image 2.5 shows up after the first render — when you need to edit, iterate, and produce a consistent set instead of one pretty image.

Use GPT Image 2.5 in the browser with credits. Every plan includes the full capability set below.
Flare when you are iterating and want speed, Sunburst when text density or edit precision matters. Start with Flare, escalate only where it falls short.
Generate from a prompt, or upload references and edit. Mask a region, describe the change, keep the rest. Both modes live in the same generator.
Draft at low or medium, ship at high, and reserve xhigh / max for print-grade detail or dense typography. Higher tiers cost more credits per image; the choice is yours per generation.
Standard 1024×1024, 1536×1024, 1024×1536, or custom WIDTHxHEIGHT up to 3840 px on a side. Pick transparent background and download real RGBA PNG or WebP.
Chinese, Japanese, Korean, and Latin scripts render in-image for posters, menus, and packaging. We have not yet published our own accuracy numbers.
Images you generate here are yours to use commercially — ads, product listings, posters, social, packaging. No per-image licensing upsell.
Choose the plan that fits. Every plan includes the full GPT Image 2.5 capability set: Flare / Sunburst, xhigh / max tiers, transparent output, 16 reference images.
Limited-Time 50% OFF
Cancel anytime.
Save $51 annually · $0.0042 per credit
Includes
Save $207 annually · $0.0033 per credit
Includes
Save $519 annually · $0.0021 per credit · lowest
Includes
Straight answers on what GPT Image 2.5 is, how Flare and Sunburst differ, and how billing works. More questions? Email support@imagesv2.ai.
The direct successor to GPT Image 2, shipped as two models: Flare for speed and Sunburst for quality.
Flare is the speed-oriented model with quality roughly comparable to GPT Image 2. Sunburst is the quality-oriented model with better editing precision and fidelity. Both cost the same credits. Draft on Flare, move to Sunburst only when Flare falls short, then drop back to Flare once the prompt is dialed in.
Same lineage, refined: up to ~50% lower latency, more natural light and texture, better reference-image preservation, more stable multi-turn edits, and new xhigh / max quality tiers. The raw-quality gain is modest — the editing and consistency improvements are the real upgrade.
No. imagesv2 gives you GPT Image 2.5 in the browser with credits. New users get free credits on signup, no credit card required.
In credits. Flare and Sunburst cost the same, scaling with the quality tier. Higher tiers and more images per run consume more credits, and the generator shows the cost before you click. See the plans above.
Up to 16 reference images per edit request, with the mask applied to the first. Four references are usually enough for identity locking.
Yes. Pick transparent background with PNG or WebP output for a real RGBA file. Sizes can be 1024×1024, 1536×1024, 1024×1536, or a custom WIDTHxHEIGHT (multiples of 16, aspect ratio up to 3:1, up to 3840 px on a side).
Free credits for new users. Pick Flare for speed or Sunburst for polish, upload references, and start editing.
