OpenAI has launched ChatGPT Images 2.5, a new generation of image creation and editing. The update is available in ChatGPT, ChatGPT Work, and Codex, while two new models are also available through the API.
The important change is more precise editing
Images 2.5 is designed to change a specific element of an image — such as clothing, a background, an object, or a visual style — while preserving the rest of the composition more reliably.
OpenAI also says the model is better at preserving characteristics from reference photos, natural lighting, and textures. Across multiple edits in the same conversation, earlier changes are more likely to remain consistent.
Flare and Sunburst
GPT-Image-2.5 Flare is optimized for speed and higher-volume generation. GPT-Image-2.5 Sunburst is designed for workflows that require more detail, precision, and control across edits.
New tools in ChatGPT
OpenAI has also added Sketch, which lets users start from a hand-drawn outline, comments on specific image regions, and templates for formats such as posters and product photography. Both models support transparent backgrounds.
Why it matters for businesses
The practical value is not limited to prettier images. Marketing and e-commerce teams often need several variations of the same asset: a different background, format, or product treatment. When a targeted edit does not destroy the rest of the image, content production becomes more controlled and repeatable.
Human review is still necessary for product accuracy, text, brand consistency, and rights to reference images.
Source: OpenAI’s official announcement.