Last April, OpenAI released a new image-generation technology that already looked like it could be a step change: Images 2, an engine that turns text prompts into near-photorealistic pictures. In the past few hours, Sam Altman’s company has shipped an even more capable version, ChatGPT Images 2.5.
The new build promises sharper detail, faster generation, more precise editing, and better tools for both creating and sharing content. According to OpenAI, Images 2.5 produces more natural lighting and richer textures, does a better job of keeping a subject’s likeness consistent, and follows edit instructions more reliably even after several rounds of revisions. Latency for image generation has also been cut by up to 50% compared with Images 2.0.
How ChatGPT’s Images 2.5 works
ChatGPT Images 2.5 is OpenAI’s latest update to its flagship image-generation tool. On top of the improvements described above, the company has introduced a new feature called Sketch, which lets users draw directly inside ChatGPT.
The sketch can then be used as a reference for image generation. OpenAI has also added templates for the most common formats, along with the ability to share content ideas with other users.
The company has reportedly already begun rolling out two separate developer-facing versions of Images 2.5 for use in third-party apps: GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst, the latter soon to be available through OpenAI’s API. Flare carries the same gains in quality, editing and speed and is the default choice for most applications, while Sunburst is built for high-quality workflows that benefit from finer control over edits.
How to try ChatGPT Images 2.5 right now
Want to try OpenAI’s new image-generation engine straight away? Open ChatGPT in your browser or app and sign in to your account. Then click the “+” button or the paperclip icon next to the text box and select “Create image.”
Alternatively, you can simply write a prompt asking the AI to generate an image. That opens the built-in editor, where you can sketch a quick draft using Sketch or upload your own photo as a reference to edit.
Once you have a first result, you can click on individual elements of the image to edit them, or type follow-up requests in the chat bar. When you’re happy with the final result, all that’s left is to open the image and download it, so you always have it on your device.
Editor’s note
This article was originally published in Italian on money.it by Pasquale Conte on September 09, 2026 as «Come funziona Images 2.5, la nuova funzionalità AI introdotta su ChatGPT». It has been translated and adapted for an international audience by the Money.it International desk.
