ChatGPT Can Now Make Animated GIFs: What Images 2.6 Does — and What OpenAI Has Not Said

 

ChatGPT can now make you an animated GIF. OpenAI launched the feature on 4 November 2026 as part of Images 2.6, its new image model for ChatGPT, Codex and the API, and it works from a written prompt or an uploaded photo. OpenAI says Images 2.6 is available globally across all subscription tiers, including free, on desktop, mobile and web. But twelve days in, the most useful thing about the feature is also the most honest: OpenAI has confirmed what it does, and has not published the limits that matter most — how many GIFs each plan gets, how long a GIF can run, or how big the file can be.

What Images 2.6 actually adds

A GIF is a short looping animation made of still pictures, called frames, shown one after another so quickly that the eye reads them as movement. The Verge, reporting on the launch, put the change plainly: ChatGPT can now create GIFs using a prompt or a photo uploaded to the chatbot, and the whole process happens inside a single chat window.

OpenAI says the update generates images 50 percent faster than before, and it splits the developer side into two models. GPT-Image-2.6 Flare is the fast model, built for iteration: at low quality, OpenAI says it can return an image in as little as six seconds, and it improves layout quality for posters and product imagery. GPT-Image-2.6 Sunburst is the detail model, built for higher-quality generation and greater consistency across edits at medium or high quality.

Marketers and creators can animate a brand mascot for a campaign, create reactions and memes for social media, or turn a product shot into an eye-catching loop. The company behind Claude's rival is moving in a similar direction with agentic tools, as our article on Claude Duet shows, but the GIF feature is the most obviously consumer-facing piece of this release.

Hands holding a smartphone showing a colourful cartoon robot animation in a bright living room
AI-generated illustration (generated 16 Nov 2026).


How to make a GIF in ChatGPT

Describe what you want, or upload a photo, and ChatGPT returns a looping animation instead of a still image. OpenAI's own example prompt is "can you make a gif of a cat dancing?" After the GIF appears, The Verge says you can change the playback speed, view the animation frame by frame, edit individual frames, and share the finished result. Frame editing is the part that matters most in practice: if one moment in a loop looks wrong — a hand in the wrong place, a face that changes between frames — you can revise that single frame rather than starting the entire idea over.

First, the rollout is gradual. OpenAI's wording is that Images 2.6 is "available globally today in ChatGPT and Codex across all subscription tiers," but The Verge describes it as rolling out to all ChatGPT and Codex users, and at least one outlet, Decrypt, could not get the GIF feature on its accounts when it tested it. Second, the feature only makes new GIFs. Reporting on the launch says you cannot upload an existing GIF file and edit it; the animation has to be created through the supported prompts and image inputs.

The uses that work best are the short ones. A GIF is a flipbook, not a video: a handful of frames that repeat. A reaction, a mascot waving, a product spinning once, a character's expression changing across six frames — these fit the format.

Creative desk with a laptop showing animation storyboard frames and a sketchbook of cartoon doodles
AI-generated illustration (generated 16 Nov 2026).


What OpenAI has not said

OpenAI has not published how many GIFs a free account can make in a day, or how many a paying account can make. It has not published a maximum GIF length or file size for the ChatGPT app, or a per-plan limit of any kind.

The developer side: GIFs in the API

Through the API, you request a GIF by setting the output format to "gif" on the image generation endpoint, and two parameters control the animation. OpenAI's API reference puts the frame count at 2 to 16 frames, with 9 as the default, and the per-frame duration at 20 to 2,000 milliseconds, with 200 as the default. Nine frames at 200 milliseconds each produce a nominal 1.8-second loop, the shape OpenAI uses in its own documentation example.

What you controlIn ChatGPTIn the API
Start from a prompt or a photoYes, per The VergeYes, generations and edits endpoints
Playback speedAdjustable after generation, per The VergeFrame duration 20–2,000 ms per frame, default 200 ms
Frame countNot published by OpenAI2–16 frames, default 9 (API reference)
Edit individual framesYes, per The VergeYes, via edits with markup guidance
Published daily or per-plan limitsNone publishedPixel budget per request; no published daily cap

OpenAI's image-generation guide at one point listed the frame count as 2 to 64, while the API reference lists 2 to 16 — and the endpoint enforces 16. Use 2 to 16, start with the default of 9, and treat the service's own validation for your account as authoritative.

Safety, and the catch with moving images

OpenAI published a system card for Images 2.6 alongside the announcement, and it is unusually candid about the new risks. The company says it added safeguards for GIF creation on top of the safety foundations from Images 2.5, and reports the share of unsafe generations not blocked fell slightly for Sunburst, from 1.92 percent to 1.72 percent, across 3,077 adversarial prompts.

The safety note also frames what this feature is not: it is not a video generator. ChatGPT's ads business is growing in parallel — OpenAI has been testing visual ads during image generation, as our article on ChatGPT picture ads explains — and a tool that turns a product shot into a looping ad creative in one chat fits that direction. The OpenAI Decisions API shows the same company pushing agents that act, not just answer; GIFs are the consumer-friendly end of the same push.

Smartphone on a cafe table showing a looping cartoon cat animation with a coffee cup nearby
AI-generated illustration (generated 16 Nov 2026).


Who should try it first

Social media creators can turn a single idea — a reaction, a mascot, a before-and-after — into a looping post without opening editing software, and frame editing means a bad frame does not waste the whole attempt. Small businesses can animate a product shot for a campaign or a seasonal greeting, the use OpenAI leads with. And designers can test character movement quickly, using a six-frame loop to check whether a walk or a wave reads before committing to full video.

Do not build a client deliverable around a GIF limit OpenAI has not published, do not promise a client a specific GIF length or file size, and do not animate real people without their consent. And if the feature has not appeared in your account yet, wait rather than hunting for a workaround: the rollout is gradual, and OpenAI says it covers all tiers, so absence today is more likely timing than a paywall. For a sense of how the same ideas are landing in rival products, Google Wallet in Gemini shows Google pushing its assistant deeper into everyday tasks in parallel.

Key Takeaways

  • ChatGPT Images 2.6, launched 4 November 2026 (not 15 November), turns a text prompt or an uploaded photo into an animated GIF in ChatGPT, Codex and the API.
  • In ChatGPT you can adjust playback speed, edit individual frames and share the result, according to The Verge; OpenAI says the feature is available globally across all subscription tiers on desktop, mobile and web.
  • Images 2.6 is also 50% faster than before, with two API models: Flare for speed (as little as six seconds at low quality) and Sunburst for detail and consistency across edits.
  • In the API, a GIF has 2–16 frames (default 9) at 20–2,000 ms per frame (default 200 ms) — OpenAI's guide once listed 2–64, but the endpoint enforces 16.
  • OpenAI has not published per-plan GIF limits, a maximum GIF length or file size in ChatGPT, or separate GIF pricing; the rollout is gradual, so the feature may not appear in every account yet.

Conclusion

ChatGPT making GIFs is a small feature with an honest shape: a prompt or a photo in, a short loop out, with the speed and the individual frames adjustable after the fact. The unconfirmed part is the limits, and OpenAI has not published them. Use the feature for what it plainly does — reactions, mascots, product loops, storyboards — and do not promise anyone a number OpenAI has not given you.

Sources

Comments