1. News
  2. Releases
  3. OpenAI · Images
ReleaseOpenAI · Images3 min read

ChatGPT can now turn a photo or a sketch into an animated GIF

ChatGPT can now make short animated GIFs from a prompt or a photo, and it can remember a saved picture so you do not have to upload it again.

By the Dashline deskSource: Basic Tutorials
ChatGPT can now turn a photo or a sketch into an animated GIF
Photo: Dennis Cortés cortes, CC0

OpenAI released ChatGPT Images 2.6 on November 4, 2026. The headline feature is simple: ChatGPT can now turn a text prompt or a photo into a short animated GIF, right in the chat. The update is available worldwide on all ChatGPT plans, including the free one.

What changed

You describe an animation or upload an image, and ChatGPT returns a short loop. You can then change the playback speed, step through it frame by frame, edit single frames and share it. In one OpenAI example, a quick owl sketch on graph paper becomes an animated character. OpenAI points to these uses:

  • Animated brand mascots
  • Product photos that loop endlessly
  • Reaction GIFs and memes
  • Animated storyboards to test movement before a video shoot

Saved reference photos

You can save a photo as a permanent reference, give it a name, and call it up in later chats with an @mention. No re-uploading. OpenAI aims this at pictures of yourself, your pets or recurring characters. OpenAI showed the feature in posts on X, but it is missing from the official blog post. That leaves it unclear how widely it has rolled out.

Faster, and more consistent

OpenAI says images now generate about 50 percent faster. It also says the model keeps people, clothing, products and brand details steadier across several rounds of edits. That targets a familiar annoyance, where a bottle's label changes by the third tweak. There are no independent tests yet, so these figures come from OpenAI alone. The company says about 4 billion images are made each week, up from about 3 billion when Images 2.5 launched in early September.

Limits and safety

  • The GIF tool is rolling out in phases, so it may not appear for everyone yet.
  • Per Mezha, ChatGPT will not process GIFs you upload yourself.
  • Images carry C2PA metadata and an invisible SynthID watermark.
  • OpenAI applies stricter limits to GIFs than to still images.

OpenAI's own system card admits GIFs bring new risks, because a problem may only show over several frames. In a test with GIFs built to be problematic, moderation caught 77.7 percent of violations. Detection reached 96.6 percent for violent content but only around 78 percent for sexual content and self-harm. OpenAI notes its safeguards do not catch everything.

For developers

Two new API models arrive: GPT-Image-2.6 Flare and GPT-Image-2.6 Sunburst. Flare is built for speed, with about six seconds for an image at low quality. Sunburst is the larger model, made for more detail. Provider fal lists GIFs of 2 to 16 frames, and about 5 U.S. cents for one high-quality 1,024 by 1,024 Flare image at launch. Check OpenAI's own documentation before relying on those details.

The takeaway

If you post images or reuse the same mascot or product photo, try the GIF tool and saved references, but expect a gradual rollout and treat the speed claim as unproven.

More articles

You should really read these.

Next step

Let’s talk