This is the ChatGPT Images 2.5 image generation model released by OpenAI. It was announced on September 8, 2026, and is available on desktop, mobile, and the web via ChatGPT, ChatGPT Work, and the Codex. OpenAI stated that over 3 billion images are generated weekly using the GPT-Image models in ChatGPT Images and the API.<\/p>\n\n\n\n
Images 2.5 improves detail, lighting, texture, and editing accuracy over Images 2.0. It explains that generation latency has been reduced by up to 501 TP3 T. It is designed to better preserve the features of people or animals in reference photos and to ensure that previous changes carry over more stably even after multiple edits.<\/p>\n\n\n\n
Product features have also been updated. Sketch is a feature that allows users to use drawings they create directly within ChatGPT as a reference for image generation. Templates enable you to start with frequently used formats, such as flyers and product photos. You can add comments to the generated images to edit specific parts, and you can also share the prompts for the finished images.<\/p>\n\n\n\n
The key point is that design work has expanded from verbal instructions to include hand-drawn sketches, templates, and comments on images. It is tailored to a method where only specific elements are changed while the rest of the composition is maintained for tasks requiring iterative revisions, such as logos, product images, campaign visuals, and social content. GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst have been added to the API.<\/p>\n\n\n\n
OpenAI stated that it also maintains safeguards. It reduces harmful results through prompts and image inspections, and continues to use C2PA metadata and invisible watermarks on generated images.<\/p>\n\n\n\n