.png){kind=logo}
OpenAI launched ChatGPT Images 2.5 on September 8, 2026, bringing a faster image engine and stronger editing tools to ChatGPT. The update rolled out to ChatGPT, ChatGPT Work, and Codex users across desktop, mobile, and web on all tiers.
OpenAI shared the news first with Axios, and the company built the release to fix a long-standing problem: AI image tools have struggled to keep faces, pets, and fine details consistent across edits. The new engine claims up to 50 percent lower generation latency than Images 2.0.
Also read: Sam Altman Debunks ChatGPT Six-Hour Shower Water Claim
The headline feature is Sketch, a tool that lets users draw directly inside a chat to guide the final image. Users can type "@Sketch" to open a canvas and rough out an idea, whether that's a room layout, a clothing concept, or something easier to draw than describe.
OpenAI also added new templates for common formats like flyers and product photos, along with the option to leave comments directly on an image for more targeted edits.
The company said editing control matters more than raw output quality this time. Earlier versions of ChatGPT Images often lost track of a subject's likeness after a few rounds of edits.
OpenAI said Images 2.5 does a better job preserving people, pets, and key details, even as users apply changes to a specific part of an image without disturbing the rest. That consistency carries across longer conversations, so earlier edits stay intact instead of degrading with each new instruction.
OpenAI also introduced two API models alongside the ChatGPT update: GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst. Flare serves as the default option for most use cases, offering higher image quality than GPT-Image-2 at half the latency.
Sunburst targets premium workflows like campaign creative and polished product imagery, where tighter editing control matters more than speed.
OpenAI said people generate more than 3 billion images every week across ChatGPT Images and the GPT-Image API models combined.
Adobe has already added the new models to Firefly, its creative AI platform, according to comments the company shared at launch. Pricing for both new API models comes in at twice the rate of GPT-Image-2.
