Release2026-09-08

OpenAI is launching ChatGPT Images 2.5 on Tuesday. The updated image engine keeps people and pets looking like themselves more reliably. It also generates pictures faster than the previous version. New tools let people sketch an object to add or mark where a change should go. Words are no longer the only way to direct an edit. Text can be typed and placed by hand. Axios tested the model on a logo, a pet portrait and a tattoo design. Adele Li, OpenAI's product lead on images, said the sketch tools aim to make people active participants in creating images. Axios notes such systems were trained on artists' work, usually without consent or payment. Generated images carry invisible watermarks.

What changed

Earlier ChatGPT image tools struggled to keep people and pets recognisable, and edits relied on text prompts.

What it unlocks

Editing a generated image by sketching, placing text, or circling the exact spot to change.

  • images generate up to 50% faster

Sources

Send this to someone who needs it

Shares the story and its sources. Nothing about you.

What does this mean for your job?

This is the story as everyone gets it. Once a week we send you the version written for your role — what changed, why it matters for the work you actually do, and one thing to try. Free while we tune it.