ChatGPT Images 2.5 adds Sketch, faster editing and API models

Just a week after releasing GPT-6 Astra, OpenAI has now launched ChatGPT Images 2.5, an update to its image generation system with sharper details, more natural lighting, richer textures, and better preservation of people, objects, and other subjects from reference images. Editing reliability has also improved across multi-turn conversations, allowing users to repeatedly modify an image with new instructions while retaining prior changes more consistently. Image generation latency is also up to 50% lower than with Images 2.0.

ChatGPT now includes a Sketch tool, which lets users draw directly in the app and use those drawings as references for generated images. The update also adds templates for common image formats and prompt sharing so others can try or adapt image generation prompts.

For developers, OpenAI is releasing GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst through the API. Flare is the default option for most applications, with higher image quality than GPT-Image-2 and 50% lower latency, while Sunburst targets premium workflows requiring tighter control over repeated edits and polished output.

Images 2.5 is rolling out across all ChatGPT tiers, ChatGPT Work, and Codex on desktop, mobile, and web.