OpenAI launches ChatGPT Images 2.6 with animated GIFs and two faster API image models
OpenAI released ChatGPT Images 2.6 on 4 November 2026 with GIF creation and reusable reference images, plus GPT-Image-2.6 Flare and Sunburst in the API.
By Emre A.Published
Tech AI LatestOpenAI released ChatGPT Images 2.6 on 4 November 2026 for every ChatGPT and Codex tier on web, desktop and mobile. Users can turn a prompt or photo into an animated GIF, change its speed and edit single frames, and keep reference images of themselves, pets or characters for reuse. Developers get two API models, GPT-Image-2.6 Flare for speed and GPT-Image-2.6 Sunburst for quality, which OpenAI says generate images 50 percent faster.
Key facts
- Announced: 4 November 2026 (OpenAI)
- Availability: ChatGPT and Codex, all plans, web, desktop and mobile
- New in ChatGPT: animated GIFs with speed control and frame editing; saved reference images
- API models:
gpt-image-2.6-flare(speed) andgpt-image-2.6-sunburst(quality and edit consistency) - Sunburst list price: $8 input, $2 cached input, $30 output per 1M image tokens
- Speed: OpenAI claims 50 percent faster generation; Flare can return an image in about six seconds
What OpenAI released
OpenAI introduced ChatGPT Images 2.6 on 4 November 2026 and made it available the same day in ChatGPT and Codex on every subscription tier, on the web, desktop and mobile. The two headline features are on the consumer side. The first is animation: users can turn a text prompt or an uploaded image into an animated GIF, adjust its playback speed, edit individual frames and share the result as a loop. The second is reference images: ChatGPT can keep pictures of the user, a pet or a recurring character so they can be reused in new images without being uploaded again.
OpenAI also says the release brings faster generation and more precise editing. Some coverage notes that GIF creation reaches accounts gradually, so not every user may see it on day one.
Two new models for developers
For the API, OpenAI shipped two models instead of one. GPT-Image-2.6 Flare, model ID gpt-image-2.6-flare, is the fast option, and OpenAI says it can generate an image in as little as six seconds. GPT-Image-2.6 Sunburst, gpt-image-2.6-sunburst, targets higher quality and greater consistency when an image is edited several times. In its announcement post OpenAI says the two models bring 50 percent faster image generation; it does not say which workloads or settings that figure was measured on, so it is a vendor claim rather than an independent benchmark.
OpenAI's model page lists Sunburst at $8 per million input tokens, $2 per million cached input tokens and $30 per million output image tokens. The Flare page shows the same $30 per million rate for image output and points to the general pricing page for the other rates. Because image models bill by tokens, the cost of a single picture depends on size and quality settings; OpenAI has not published a flat per-image price for either model.
Where this leaves the image model race
Images 2.6 follows ChatGPT Images 2.0 and the gpt-image-2 API model, which OpenAI launched in April 2026. Splitting the API into a fast model and a quality model mirrors what text model vendors already do and gives developers a cheaper path for high-volume work such as thumbnails or drafts, with the slower model kept for final assets.
Rivals have moved in the same weeks. Google cut prices with Nano Banana 2.1 in October, and Microsoft released MAI-Image-2.7, an editing model, in preview on 27 October. GIF output is the clearest point of difference for OpenAI: it moves ChatGPT from still images towards short animation without asking users to switch to a video model. It also puts more generated motion content into chat apps and social feeds, which raises the usual questions about labelling and likeness, given that reference images make it easier to place real people in new scenes.
What to watch next
Developers should check whether GIF output is exposed in the API for both models and with what limits on frame count and length, and whether OpenAI adds a batch discount for the 2.6 models. OpenAI has not published benchmark results for Flare or Sunburst against other image models, so independent leaderboards such as Arena and Artificial Analysis will be the first outside test. The safety card or system notes for the release, including how reference images of real people are handled, are not stated in the announcement. Other model launches are tracked in the AI models section and the model release tracker.
What changed
| Before | ChatGPT Images produced still images; the API offered a single current image model, gpt-image-2 |
|---|---|
| After | ChatGPT Images 2.6 adds animated GIFs and saved reference images; the API adds gpt-image-2.6-flare and gpt-image-2.6-sunburst |
| Who it affects | ChatGPT and Codex users on all plans; developers building image features on the OpenAI API |
Questions people ask
- Can ChatGPT make GIFs?
- Yes. Since ChatGPT Images 2.6 on 4 November 2026, ChatGPT can turn a prompt or image into an animated GIF and let users change its speed and edit single frames. OpenAI says it is available on all plans.
- What is the difference between GPT-Image-2.6 Flare and Sunburst?
- Flare is the speed model, which OpenAI says can return an image in about six seconds. Sunburst is aimed at higher quality and consistency across repeated edits.
- How much does GPT-Image-2.6 Sunburst cost?
- OpenAI's model page lists $8 per 1M input tokens, $2 per 1M cached input tokens and $30 per 1M output image tokens.
Sources
- Images, upgraded: introducing ChatGPT Images 2.6 (OpenAI, 4 Nov 2026)
- GPT-Image-2.6 Sunburst model (OpenAI, 4 Nov 2026)
- GPT-Image-2.6 Flare model (OpenAI, 4 Nov 2026)
- OpenAI announcement post on X (OpenAI, 4 Nov 2026)
- ChatGPT can now make animated GIFs with OpenAI's Images 2.6 (The Next Web, 4 Nov 2026)
Reference pages
- GPT-Image-2.6 model
- OpenAI company
- ChatGPT tool
- Codex tool
Spotted an error? See the corrections policy and tell the editor. Facts in this story come from the sources listed above; vendor claims are attributed.
Comments
Loading comments.





