OpenAI added Sketch to ChatGPT — draw a rough guide in the chat with @Sketch and the model renders it into a detailed image. It shipped with ChatGPT Images 2.5 on September 8, 2026.
2026-09-08 · by Moe Ameen
OpenAI added a feature called Sketch to ChatGPT, letting you draw a rough guide directly in the chat and have the model turn it into a detailed image. You invoke it by typing @Sketch, doodle the layout of a scene — the shape of a room, the contour of an outfit, the arrangement of objects in a frame — and add a text description of the style and details you want. ChatGPT uses the drawing as a compositional reference rather than a literal image, so a crude scribble is enough to steer where things go. OpenAI's framing is that you do not need to be an artist; the sketch is just a faster way to communicate layout than a paragraph of prompt.
Sketch arrived as part of ChatGPT Images 2.5, OpenAI's new image model announced September 8, 2026, which also brought Templates (prebuilt starting points for formats like posters and merch), the ability to leave comments directly on a generated image for focused edits, and up to 50% faster generation than the prior Images 2.0. Sketch is rolling out with the model to all ChatGPT, ChatGPT Work, and Codex users across desktop, mobile, and web.
The appeal is control. Text prompts are good at describing content and style but weak at pinning down composition — where a subject sits, how elements are arranged, what the framing is. A rough drawing communicates that instantly. Sketch closes the gap between having a layout in your head and getting the model to honor it, without learning prompt tricks or fighting the generator across a dozen retries.
Sketch is a control surface for one image; the interesting move for a creator is treating that controlled image as the seed of a whole drop. Sketch a thumbnail or a hero layout, describe the style, let ChatGPT render it — then bring that image into [Kompozy](/) and fan it into a set: a brand-exact [carousel](/glossary/hyperframes) built with HyperFrames, a Quote Graphic pulling the strongest line, and Photo Posts with per-platform copy written in your voice through the [Persona Brief](/glossary/persona-brief). One sketched idea becomes a coordinated, scheduled drop across nine platforms plus blog and email, instead of a single file you still have to caption and post by hand.
There is also a fast, timely play. "ChatGPT can turn your drawings into images now" is a high-intent search this week, and one point of view about it can become a week of content in Kompozy — a short explainer, a carousel breaking down how Sketch works, captioned clips, and platform-native posts, all scheduled with autopilot and a review step. The parallel is almost too neat: Sketch takes a rough gesture and makes it a finished image; Kompozy takes a finished image and makes it a finished, published feed. For the full picture of the model Sketch shipped inside, see our [ChatGPT Images 2.5 launch coverage](/news/chatgpt-images-2-5-launch).
It is a feature in ChatGPT that turns a rough drawing into a detailed image. You type @Sketch, draw a rough guide for the layout in the chat, add a text description of the style and details, and the model renders a finished image using your sketch as a compositional reference. It shipped with ChatGPT Images 2.5 on September 8, 2026.
Type @Sketch in the chat to open the drawing canvas, sketch the rough layout you want — a room, an outfit shape, the arrangement of a scene — and describe the style and any details in text. ChatGPT treats the drawing as a guide for composition rather than a literal image and generates a detailed result. You do not need drawing skill; a crude scribble is enough to steer placement.
Sketch rolled out with ChatGPT Images 2.5 to all ChatGPT, ChatGPT Work, and Codex users across desktop, mobile, and web. Paid tiers generally get faster and higher-volume image creation, and some plan-level image caps are not officially published, so confirm current limits on OpenAI's pricing page.
No. Sketch controls the composition of a single generated image; it does not caption for social, turn one image into a branded multi-format set, or publish anywhere. That distribution work is separate — a content engine like Kompozy takes a generated image and fans it into a carousel, quote card, and per-platform posts, then schedules them across nine platforms.