OpenAI has introduced ChatGPT Images 2.5, a model update that includes a new Sketch tool. According to the company, the feature lets users draw directly in the chat window and then have their rough doodles converted into polished, more detailed images. The goal is to make image generation feel more personal and aligned with the user's original idea.
ZDNET's tutorial confirms the same workflow: you sketch inside ChatGPT, and the tool refines your drawing into a final image. The two sources agree on the core capability, though they differ in emphasis. OpenAI's announcement frames it as part of a broader model upgrade, while ZDNET provides practical steps for trying the feature yourself.
If you have access to ChatGPT, the Sketch tool is available now. You can start with a simple drawing, add a text prompt if you like, and let the model handle the rest. The result is a more polished version of your original sketch, which may be useful for quick concept art, storyboarding, or just experimenting with AI-assisted creativity.