ChatGPT's image tool gets faster, with better editing and a Sketch mode

OpenAI has released ChatGPT Images 2.5, a new version of the image maker inside ChatGPT. The company says it is faster, produces sharper pictures and lets you change one part of an image without disturbing the rest. It also adds a Sketch tool, where you draw a rough layout and ChatGPT turns it into a finished image.

What changed

OpenAI says the new model gives sharper details, more natural lighting and richer textures. It also says the model is better at keeping people and subjects looking the same when you start from a reference image. It is meant to follow complex instructions more closely, including the layout and visual style you ask for.

On speed, OpenAI claims Images 2.5 can cut waiting time by up to 50 percent compared with Images 2.0. These are the company's own claims, and the material does not include independent tests.

Editing and new tools

Editing is the main focus. You can change an object, the background or the text in a picture while the main subject and overall layout stay put. OpenAI also says changes made in earlier steps should hold up better as you keep editing, so quality and details are less likely to shift in a long session.

• Sketch: draw a rough layout or doodle in ChatGPT and add a text prompt to get a finished image.
• Templates: start from formats such as posters, flyers, merchandise and product photos, then add your own details.
• Comments: place notes directly on an image to ask for focused edits.
• Prompt sharing: pass your prompts to other users.

Who it affects

Small business owners, marketers and anyone who makes social posts, ads or flyers are the most likely to notice. Templates mean you do not have to start from a blank prompt. Sketch helps if you know the layout you want but find it hard to describe in words.

OpenAI also says users now create over 3 billion images every week across ChatGPT Images and its GPT-Image models, which developers can reach through the API. That is a way for apps to connect to the technology.

Where to find it

Images 2.5 is available in ChatGPT, ChatGPT Work and Codex, on desktop, mobile and the web. Developers can use two models through the API, called GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst.

What is not known yet

The material does not say what Images 2.5 costs, whether limits differ by plan, or how the speed and editing claims hold up in everyday use. A small, low-stakes task is a sensible way to test it first.