OpenAI · 2026-09-08 · major
ChatGPT Images 2.5 — OpenAI's image model adds Sketch and two API tiers
ChatGPT Images 2.5 is OpenAI's new image model, with sharper detail, better likeness from reference photos and up to 50% lower generation latency than Images 2.0. Two API models ship with it: GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst.

OpenAI's Images 2.5 keeps the subject of a reference photo intact while you edit everything around it.
Key specs
| Latency vs images 2.0 | up to 50% lower |
|---|---|
| Images made weekly | 3B+ |
Quick facts
| Maker | OpenAI |
|---|---|
| API models | gpt-image-2.5-flare and gpt-image-2.5-sunburst |
| Availability | ChatGPT, ChatGPT Work and Codex on desktop, mobile and web |
| New in ChatGPT | Sketch, templates, comments on images, shareable prompts |
| Provenance | C2PA metadata plus invisible watermarking |
| Partners at launch | Adobe Firefly, Higgsfield AI, Manus |
What is it?
ChatGPT Images 2.5 adds a drawing tool called Sketch, image templates, comments placed directly on an image, and shareable prompts so other people can rerun your idea on their own photos. The model underneath is new too: OpenAI calls Images 2.5 state of the art, with more natural lighting, richer textures, and subjects from a reference photo that stay recognizable across new settings and styles.
How does it work?
Editing is where the model changes most. OpenAI says Images 2.5 edits only what you asked for and leaves the rest of the frame alone, and that it holds earlier changes steady through a long back-and-forth instead of degrading quality with each turn. It also handles more complex layouts, including transparent backgrounds, and sticks closer to a stated visual style as instructions get more specific.
Why does it matter?
For developers the split into two API models is the practical part: GPT-Image-2.5 Flare is the default and produces higher-quality images than GPT-Image-2 at 50% lower latency, while GPT-Image-2.5 Sunburst trades generation time for tighter control on premium work. Adobe has added the new models to Firefly, and Manus reports Flare running two to four times faster than GPT-Image-2 in its own evaluations.
Who is it for?
designers, marketers and app developers
Frequently asked questions
- Is ChatGPT Images 2.5 available on the free plan?
- OpenAI says Images 2.5 is rolling out to ChatGPT, ChatGPT Work and Codex users across all tiers, on desktop, mobile and web. The announcement does not publish per-plan generation limits. Developers reach the model separately through the Images API using the two new models, GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst.
- What is the difference between Flare and Sunburst?
- GPT-Image-2.5 Flare is OpenAI's default choice for most applications — creator and social content, product experiences, visual search, rapid prototyping and high-volume generation. GPT-Image-2.5 Sunburst is built for premium visual workflows that need tighter control across edits, such as campaign creative or polished product imagery, and takes longer to generate.
- How can you tell an image came from ChatGPT Images 2.5?
- OpenAI attaches C2PA metadata and an invisible watermark to images made with its tools. Images 2.5 also builds on OpenAI's existing safeguards with checks on both prompts and generated images, and OpenAI points to the model's system card for the details of its evaluations and safety approach.
- What is Sketch in ChatGPT?
- Sketch is a new ChatGPT feature that lets you draw directly in the app and use the drawing as a visual guide for the final image. OpenAI's examples include sketching a room layout or the outline of an outfit; you add a description of the style and any other details, and ChatGPT turns the rough drawing into a finished image.
Try it
Call gpt-image-2.5-flare on the OpenAI Images API