OpenAI just released ChatGPT Images 2.5: faster, sharper, and a lot more controllable.

OpenAI just released ChatGPT Images 2.5: faster, sharper, and a lot more controllable.

ChatGPT Images 2.5 feels less like “type a prompt and hope” and more like an actual creative workspace.

OpenAI has launched ChatGPT Images 2.5, cutting generation latency by up to 50% while improving reference-image fidelity, natural lighting, textures, and consistency across repeated edits.

New editing controls also let you target specific areas of an image instead of regenerating the whole thing.

Here’s the real story:

The biggest upgrade may be control.

The new Sketch feature lets you draw what you want directly inside ChatGPT, while image comments let you point at specific elements and tell ChatGPT exactly what to change.

Templates also give creators structured starting points for things like posters and product photography.

For developers, GPT-Image-2.5 Flare prioritizes faster generation while GPT-Image-2.5 Sunburst is aimed at higher-precision work where quality matters more than speed.

OpenAI says people are already creating more than 3 billion images per week across ChatGPT Images and its image API models.

AI image generation is starting to move beyond prompting. The next battle is who builds the best creative workflow around the model.