gpt-image-2.5-vip reverse series is live and callable in the Default group
Three new models: gpt-image-2.5-vip (alias of gpt-image-2.5-sunburst-vip), gpt-image-2.5-flare-vip (speed-first) and gpt-image-2.5-sunburst-vip (quality- and editing-first). They run on the same Adobe reverse line as gpt-image-2-vip (Firefly, a high-quality GPT-Image 2.5 reverse line, not low-quality upscaling) with the same $0.03 per-image price, groups, endpoints and call format; just swap the model name.
A three-arm comparison on the same channel and token before launch showed all three accept quality up to high (xhigh / max unavailable), support background: "transparent", and lock all 30 size presets pixel-exact. Note that 2.5 high only equals gpt-image-2-vip medium, so do not carry quality over verbatim when migrating; mask is whole-image regeneration on all three, so use the official models for precise inpainting.
On the ChatGPT web reverse line, gpt-image-2.5-all ships alongside: the ChatGPT web app has moved to Images 2.5 as a whole, so gpt-image-2-all already produces 2.5 images; the new name simply makes the upgrade explicit, and the two names share price and behavior.
The docs sidebar groups are now “GPT-Image-2.5-VIP Image Gen” and “GPT-Image-2.5-All Image Gen”; the old model pages and URLs are unchanged.
📖 Row-by-row comparison of the three -vip models: GPT-Image-2.5-VIP Overview · Official vs reverse selection: comparison page
← Back to Live Updates · 📚 Monthly archive