AI tools · Checked 5 November 2026, 00:10 UTC.
OpenAI announced Images 2.6 on 4 November 2026, adding animated GIFs and new editing controls. OpenAI says the release is available worldwide in ChatGPT and Codex across subscription tiers on desktop, mobile and web. Check the availability and usage limits in your account.
Developers should check the endpoint
OpenAI positions Flare for fast iteration and Sunburst for detailed visual work in the API.
The generation guide specifies GIF output via POST /v1/images/generations with output_format="gif"; the edit endpoint cannot return GIFs. Choose 2–64 frames and 20–2000 ms per frame, rounded to 10 ms; defaults are nine frames at 200 ms. Size is auto or positive width/height multiples of 16. Frame count, dimensions and animation count must fit size and total-pixel limits. GIF output supports neither streaming nor partial images, custom output_compression or prompt_rewriter.
Budget for the output
Standard API pricing for either model, per million tokens: US$30 for image output, US$8 for image input and US$5 for text input. Cached-input rates are US$2 for images and US$1.25 for text. These are token rates, not a fixed price per GIF. Check measured usage before estimating a batch.
A small test before publishing
- Use an original, non-sensitive image and one simple movement.
- Inspect every frame for changing text, logos or unintended details, then watch the transition back to the first frame.
- Test the downloaded file where readers will actually see it. Keep a static alternative for readers who prefer less motion.
We have not tested the rollout or benchmarked output quality. These are documentation-based checks; review the generated file before using it in a campaign or interface.