OpenAI ships ChatGPT Images 2.6, and it can reportedly make GIFs

What we actually know

As Thurrott.com reported on November 5, 2026, OpenAI has put out ChatGPT Images 2.6. It is a new numbered release of the image generation feature built into ChatGPT, the part that turns a typed description into a picture.

The one concrete change we have seen is that the feature can reportedly now create GIFs. A GIF is a short, silent image that loops, the kind often used for quick reactions or small animated graphics. That is a different output from the still pictures the feature is known for.

That is close to the full list. The report we have is brief, and the GIF claim is labeled as reported, not confirmed by us.

What is still missing

Several basic questions are open. We do not know what else changed in 2.6, such as whether the pictures look better, handle text more cleanly or follow instructions more closely. A version number alone says an update happened, not what improved.

We also do not know who gets it. It is unclear whether every ChatGPT user can use it or only people on certain plans. The price is not stated, and neither are any limits on how many images or GIFs you can make.

Until OpenAI or a hands-on review fills these in, treat every claim about 2.6 as provisional. That includes the GIF claim, which comes from a thinly detailed report and which we have not tried ourselves.

Why a small update can still matter

Incremental image updates are easy to wave off. Nobody rewrites their workflow because a feature went from one version to the next. But these updates can quietly move a tool from fun to usable for real work.

Think of the everyday jobs where an image has to be good enough to ship: a slide that needs an illustration, a social graphic, a quick mockup of an idea. If the generator gets a little better at any of those, the time saved adds up over a month, even if no single image looks dramatically different.

An output format like GIFs also changes what you might ask for. Instead of a static picture, you could request something with a little motion. Whether the results are good enough to use is exactly what we cannot say yet.

A fair way to test it

The practical approach is a quick, honest comparison, not a switch. Pick three or four prompts, meaning the written instructions you give the tool, from work you actually do. Use the same ones you would normally rely on, not showpiece examples.

• Include a prompt that needs readable text. Text inside images is a common weak spot for image generators, so it is a revealing check.
• Include one that needs a consistent look. If you make several images that must match, see whether the style holds together.
• Try a GIF request. If the option appears for you, note how well it follows your instructions.

Then compare the results against what you were getting before. If 2.6 is clearly better on your tasks, change your habits. If it is a wash, you have lost ten minutes and learned something.

What to watch next

The gaps above are the things to follow. Look for OpenAI's own description of what changed in 2.6, or hands-on reviews that go beyond the headline. Those would tell us whether the GIF feature is the main story or just one item on a longer list.

Also watch for plan availability and usage limits. A feature that is available to everyone is a different story from one limited to paying users, and limits decide whether you could rely on it for regular work.

For now, the honest summary is short: a new version exists, it reportedly makes GIFs, and the rest has to be checked by trying it.