OpenAI launches ChatGPT Images 2.5 with Flare and Sunburst

OpenAI launched ChatGPT Images 2.5 on 8 September, with GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst available in the API. Flare is positioned as the default for everyday generation; Sunburst targets workflows where precise editing matters most.

OpenAI reports improvements in focused edits, consistency across multiple edits and visual instruction following. The release also adds Sketch and creative templates in ChatGPT. Product features in ChatGPT should be distinguished from the API models themselves.

The native identifiers are gpt-image-2.5-flare and gpt-image-2.5-sunburst. The release changelog lists support through the Image API and the Responses API image-generation tool. Consult current pricing for the chosen size, quality and operation rather than treating every generated image as the same unit of work.

Keep an approved original image beside every edited version. Build a small review set with a product photograph, a transparent logo, a layout containing text and a complex background. Ask for one bounded change at a time, then inspect the areas that should have stayed unchanged. An attractive result can still introduce the wrong product detail or alter a client's branding.

For repeatable production work, save the prompt, model ID, requested quality and output dimensions with the asset. Review text at its final display size, and check transparency against both light and dark backgrounds. Human approval remains useful before a generated asset enters a public campaign. These are AZ Labs workflow recommendations; we have not claimed an independent quality or speed measurement for either new model.