OpenAI has announced ChatGPT Images 2.5, a major upgrade to ChatGPT’s image generation capabilities. The standout highlight is the Sketch feature, which allows users to draw directly inside ChatGPT and use that drawing as a reference to generate a complete image.
Using it is fairly simple: type “@Sketch” into the chat box, and a drawing window will appear. Users can sketch anything—from a room layout or an outfit to a simple doodle—and then describe their desired style and details. ChatGPT will turn that sketch into a detailed image.
According to reporter Jay Peters of The Verge, he tried drawing a cat using a computer mouse, and ChatGPT turned it into a realistic cat photo based on his description. He also tried the inline commenting feature on images: clicking on the cat’s eye area and asking to change the color from blue to green, which ChatGPT performed accurately.
Beyond Sketch, OpenAI stated that this upgrade significantly improves image quality: sharper details, more natural lighting, richer textures, and crucially, better preservation of characteristics from reference images. Image generation time has been reduced by up to 50% compared to Images 2.0, an important enhancement for those who need multiple iterations to achieve their desired result.
Another new feature is Template, which allows users to start from popular formats such as posters or product photos instead of from scratch. Along with this is the ability to share prompts, enabling other users to try similar ideas with their own images and details.
OpenAI also introduced two new API models for developers:
- GPT-Image-2.5 Flare is the default option, bringing improvements in quality, editing, and speed.
- GPT-Image-2.5 Sunburst is a longer-running version designed for tasks requiring higher precision in detailed edits.
According to OpenAI, over 3 billion images are currently generated each week through ChatGPT Images and GPT-Image API models. This figure shows that AI image generation has become one of the most popular applications of generative AI, far exceeding what many users expected just a few years ago.
Images 2.5 is available to all ChatGPT, ChatGPT Work, and Codex users across desktop, mobile, and web. This is the next step following Images 2, the previous upgrade that added 2K output, multiple aspect ratios, web research support, and the ability to process non-Latin languages.
For regular users, Sketch could be a feature that changes how they interact with image generation AI. Instead of having to write long and detailed prompts, they can now quickly draw a sketch and let the AI handle the rest. This is particularly useful for people who are not adept at expressing themselves in language but have a clear idea of the desired image.