OpenAI Updates Images 2.6 With GIF Generation and Saved References

OpenAI has introduced Images 2.6, an updated image generation model that brings animated GIF creation to ChatGPT alongside traditional image generation.Users can create animations from text prompts or uploaded sketches, adjust playback speed, and share the resulting GIFs. The feature is gradually rolling out to ChatGPT users.Image editing has also improved. The model is better at preserving character appearances, product details, and other visual elements across multiple editing rounds. According to OpenAI, image generation is now 50% faster than with the previous version.Another addition is saved reference images. For example, users can upload a photo of their cat, save it as "Murzik," and later request an illustration of @Murzik in a specific style in a different conversation. ChatGPT retrieves the saved reference automatically, eliminating the need to upload the image again.For developers, OpenAI has introduced two models through its API:

GPT-Image-2.6 Flare is optimized for fast image generation and editing. At low quality settings, it can generate an image in as little as six seconds.
GPT-Image-2.6 Sunburst is designed for more detailed visual work. At medium and high quality settings, it delivers higher image quality and improved consistency across successive edits.

The API also introduces additional controls, including markup-guided editing to specify which areas of an image should be modified, SVG export, and a complexity parameter that determines how much planning the model applies to a request.Images 2.6 is now available in ChatGPT and Codex, while Flare and Sunburst are accessible through the API.You can also try GPT Image 2.6 on BotHub for AI image generation.