OpenAI launched ChatGPT Images 2.6 on Wednesday, introducing native animated GIF creation, faster generation speeds and two developer models. The update rolled out globally across all consumer tiers, including ChatGPT Pro accounts, the Codex programming tool and OpenAI’s developer platform.
Key points
- ChatGPT Images 2.6 creates animated GIFs from written descriptions, existing images or rough sketches.
- The underlying model generates images 50 per cent faster than Images 2.5, with an API low-quality preset running in six seconds.
- OpenAI introduced two API models, Flare and Sunburst, which add vector export, markup edits and complexity controls.
- The release marks OpenAI’s first moving visual tool since closing its Sora video service earlier this year.
How does ChatGPT Images 2.6 generate animated GIFs?
The updated tool allows users to make animated GIFs by describing what they want in a text prompt or by uploading an existing still image to animate. Users can also supply a rough hand drawing, which the AI model develops into a finished visual before animating the sequence.
Control over the output goes beyond text prompts. The Verge reported that the ChatGPT interface allows users to adjust animation playback speeds, view individual frames and edit single frames directly to refine the animation.
In a since-deleted post on X, OpenAI stated that users would also be able to save uploaded pictures within the system to reuse them as prompts in future sessions.
What are the GPT-Image-2.6 Flare and Sunburst API models?
Images 2.6 cuts standard image generation time by 50 per cent compared to the Images 2.5 model it replaces. On its developer platform, OpenAI introduced two API releases: the speed-focused GPT-Image-2.6 Flare model and the GPT-Image-2.6 Sunburst model.
Flare is aimed at developers who need low latency. OpenAI said the model can generate an image in approximately six seconds when users choose its low-quality preset.
Both Flare and Sunburst add new functional controls for software developers. MarkTechPost reported that the two API models support animated GIF output, vector graphics through SVG export, markup edits and an adjustable complexity parameter that governs the level of detail.
Why is OpenAI releasing GIFs instead of full AI video?
OpenAI stated that the GIF capability in Images 2.6 is not an AI video generator. The company instead pitched the format as an intermediate tool for creators to assemble animated storyboards and “explore character movement before video production.”
The release follows the shutdown of Sora, OpenAI’s standalone generative video application, earlier this year. The Wall Street Journal reported that Sora attracted few users and cost the company roughly $1 million a day to operate.
Since discontinuing Sora, OpenAI has concentrated its visual AI work on still images and user interfaces. Last month, the company released Intelligent UI, a ChatGPT feature that lets users generate interactive visual interfaces inside chat sessions.
