ChatGPT Sketch Turns Mouse Doodles Into Detailed AI Images

OpenAI has officially rolled out ChatGPT Images 2.5 on Tuesday, introducing a major functional upgrade that redefines how users communicate visual intent to language models. The headline addition is a feature called Sketch, which allows creators to bypass text-only prompting by drawing crude shapes directly inside the chat interface. According to The Verge, users can activate this canvas by typing @Sketch into the prompt box, pulling up a dedicated window where even the roughest mouse-drawn shapes can be paired with descriptive text instructions to generate polished visual assets. This lowers the friction for non-designers who struggle to articulate complex layouts through text alone, turning rough outlines into fully realized graphics.

How Does the ChatGPT Sketch Feature Work?

The ChatGPT Sketch tool operates via a dedicated modal window that launches when typing @Sketch into the active text field, giving users a digital canvas to draft rough outlines, shapes, or stick figures using a mouse or trackpad. Once the drawing is complete and submitted alongside a text prompt, OpenAI's underlying image generation architecture interprets both structural spatial lines and semantic intent to synthesize a cohesive, high-resolution output. Early hands-on testing demonstrates that the system successfully translates poorly proportioned mouse sketches into detailed photographs or illustrations based strictly on user parameters. This multimodal input method bridges the gap between abstract conceptualization and precise visual generation, offering a tactile alternative to pure prompt engineering across various professional workflows.

New Annotation Tools Enhance Image Editing

Alongside the Sketch canvas, the ChatGPT Images 2.5 release introduces granular editing capabilities that allow users to leave direct comments on specific areas of generated images. Instead of regenerating an entire visual asset due to a localized flaw, creators can now point to exact coordinates within the frame to request targeted adjustments. This capability brings generative AI closer to traditional design software workflows, where iterative refinement is standard practice. Enterprise teams testing generative visual pipelines often find all-or-nothing regeneration frustrating when minor details fail, making localized annotation a welcome addition for professional design tasks.

• OpenAI launched ChatGPT Images 2.5 on Tuesday with advanced visual generation tools.
• The new @Sketch command opens an interactive drawing window inside the chat interface.
• Users can combine rough mouse doodles with text instructions to guide final image output.
• Direct image annotation lets creators leave comments on specific sections for targeted edits.

"With a new feature called Sketch, you can just draw something right inside ChatGPT and then tell ChatGPT how you want it to make an image based on that sketch."

What to Watch Next

As OpenAI rolls out ChatGPT Images 2.5 and its companion Sketch tool to the broader user base, enterprise tech watchers should monitor three critical adoption signals in the coming weeks. First, watch for enterprise workspace integration announcements regarding how administrative controls handle user-generated sketches and proprietary data input. Second, observe whether OpenAI expands the Sketch interface to support stylus inputs and tablet touchscreens natively for professional digital artists. Third, track user feedback on how effectively the system maintains layout consistency when handling complex multi-element drawings versus simple object outlines.