OpenAI announced ChatGPT Images 2.5 on September 8, 2026. The Sketch feature introduced with the update allows users to use a simple doodle they draw within ChatGPT as part of a visual prompt. A drawing can be made in the window that opens when @Sketch is typed into the chat box; the visual is then generated based on that drawing. In the test, a poorly drawn cat doodle made with a computer mouse was turned into a realistic photograph in line with the instruction. Users can leave comments directly on the areas of the images they want changed; using this method, the cat’s eyes were changed from blue to green. OpenAI said the release offers more natural lighting and richer textures, follows editing instructions better across multiple rounds, and reduces latency by up to 50% compared with Images 2.0. ChatGPT Images 2.5 was made available to ChatGPT, ChatGPT Work and Codex users on desktop, mobile and the web.
Why it matters
This update allows users to express their ideas more directly by reducing the need to rely solely on written descriptions for image generation. The ability to incorporate a simple drawing into the generation process expands the scope of control, especially for users who have difficulty describing a composition or its basic form in words. Changes made through regional comments also make progressive editing of an image, rather than one-off generation, part of the process within ChatGPT. However, the extent to which OpenAI’s claims about speed and instruction following apply across different drawing, prompt and editing examples remains an open question; the reporter’s test in the announcement provides only a limited example on this matter.
Background
ChatGPT is not a new name in the FikirPilot archive: we have published 18 news stories mentioning this name in the past 90 days; the most recent was dated September 12, 2026.