ChatGPT Images 2.6 Makes Animated GIFs: What Is Reported and What You Can Check

ChatGPT can now make animated GIFs, according to The Verge. The feature arrives with a new image update that The Verge calls Images 2.6, plus a claim that users now generate over 4 billion images per week.

We could not open OpenAI's own announcement when we wrote this. So this post separates what The Verge reports from what we could check in OpenAI's public documentation, and shows how to test the feature yourself.

TL;DR: what is reported and what is checked

What does the GIF feature do?

The Verge says ChatGPT "can now create GIFs using a prompt or a photo uploaded to the AI chatbot." The example OpenAI shared asks for a cat dancing. The result comes with a playback speed control and a view of individual frames, which you can edit.

Before this change, the report says, ChatGPT could not natively generate a GIF in a single step. People used workarounds. A common one was to ask for a numbered sequence of still frames and then have ChatGPT stitch them into an animated file. Our own guide to stop-motion animation with GPT-Image 2.5 shows that frame-by-frame method.

The new part is the packaging. One request returns a loop, and the loop is editable inside the chat. That matters for people who make reaction images, product demos, slide animations or small looping banners.

What is Images 2.6, and how does it relate to 2.5?

OpenAI's image line moved fast in 2026. ChatGPT Images 2 arrived in April. Then Images 2.5 with the Flare and Sunburst models launched in September, and OpenAI said at that time that over 3 billion images were generated each week.

The names are the key clue. In OpenAI's image generation guide, the API uses gpt-image-2.5-sunburst and gpt-image-2.5-flare. The guide says to choose Sunburst when editing precision matters most and Flare for fast, high-quality everyday generation.

The Verge reports that Images 2.6 carries the same two names forward: GPT-Image-2.6 Flare and Sunburst, in the API, "for faster and higher-quality image generation." Earlier in the autumn we tracked the unannounced model names that appeared in OpenAI's API spec; see our post on the unannounced OpenAI API image models. The Verge report is the first public sign we found that a 2.6 generation is launching.

When we read the guide, it still named only the 2.5 models. That is not a contradiction. Docs often lag a launch by hours. It does mean you should not hard-code a 2.6 model ID in production until it appears in the docs or in your model list call.

The 4 billion figure: a usage claim, not a benchmark

The Verge says OpenAI "says users now generate over 4 billion images per week across ChatGPT and its API," up from 3 billion reported in September. That is a one-third rise in about two months, if both figures are comparable.

Treat this as a company-reported usage number. It does not say how many are GIFs, how many are edits, or how many come from the API. It also does not tell you whether the 3 billion and 4 billion counts use the same definition. We covered a similar pattern in our post on ChatGPT usage claims, claimed vs verified: big usage numbers are useful for scale, weak for comparison.

A short timeline of OpenAI's image releases in 2026

The pattern is a new image generation roughly every two months, each with a speed claim and a usage number. That pace is why docs, model lists and third-party tests often disagree for a few days after a launch. It is also why a claimed-versus-verified habit pays off: record the model ID you actually called, the date, and the output, so you can compare later.

What is still unknown?

• Limits. How long can a GIF be, how many frames, what resolution, and what file size? The report does not say.
• Plan access. The Verge says Images 2.6 is rolling out to all ChatGPT and Codex users. Rollouts are staged, so you may not see it today.
• API behavior. The OpenAI image API has long offered still-image formats. Whether the 2.6 API models output GIFs, or whether GIF creation is a ChatGPT-only layer, is not stated in what we read.
• Pricing. No price for the 2.6 API models appears in the report.
• Speed. The Verge says generation is faster. No latency number is given. For reference, OpenAI's developer account said Flare cut latency by 50% against GPT-Image-2 at the 2.5 launch.
• Safety and likeness. Animating an uploaded photo raises the usual questions about people's likeness. We have not seen a policy statement for the GIF feature.

How to test the GIF feature in 10 minutes

1. Update the app. Use the web, desktop or mobile app and check for the update.
2. Start simple. Ask for a short looping GIF of one object on a plain background. Complex scenes are harder to judge.
3. Open the speed control. Change the playback speed and check that the loop stays smooth.
4. Use the frame view. Edit one frame, then replay. Look for flicker or style drift between frames.
5. Try a photo. Upload a photo you own and ask for a small motion, such as blinking or a drifting cloud. Do not use other people's photos.
6. Export and inspect. Download the file, open it in a browser, and note file size and loop behavior. Large GIFs load slowly on the web.
7. Compare. Run the same prompt with the old frame-stitching method from our stop-motion guide and see which gives steadier results.

Why this matters for builders

GIF generation is a small feature with a clear use. Designers need quick loops for docs, social posts and release notes. Support teams use short animations to show a click path. Marketers need cheap variants.

For developers, the open question is the API. If the 2.6 API models return animated output, tools that now assemble frames by hand could drop that code. If they do not, ChatGPT's GIF editor is a consumer feature and your pipeline stays the same. Wait for the docs before you plan around it.

OpenAI is also pushing images deeper into products. We have tracked ChatGPT visual ads and image generation tests, and rivals such as Microsoft MAI-Image-2.6-Flash compete on speed. A faster default model plus easy animation raises the pressure on those tools.

What readers are asking

Is this the same as Sora? No. The report describes GIF creation inside the image feature. It does not describe video generation.

Will it be free? The Verge says it rolls out to all ChatGPT users. Free-tier limits for images have changed before, so check your plan's image allowance.

Does it replace GIF editors? It covers quick creation and light editing. Fine control over timing, color palettes and optimization still needs a dedicated editor.

Are the numbers independent? No. The 4 billion figure and the model names come from OpenAI through The Verge.

Bottom line

The reported change is simple and useful: ask ChatGPT for a GIF, adjust speed, and edit frames. The surrounding claims, including Images 2.6, the new Flare and Sunburst API models and 4 billion images a week, all trace to OpenAI through one outlet. We expect the docs and OpenAI's post to confirm or correct the details within days. We will update this post if they differ.

Details reflect The Verge's November 4, 2026 report and OpenAI's image generation guide as read that day. Model names, limits and availability can change.

Related reading

• ChatGPT Images 2.5: Flare and Sunburst launch
• How to make stop-motion animation with GPT-Image 2.5
• ChatGPT Images 2 and GPT-Image-2
• Unannounced OpenAI API image models: what we know
• ChatGPT visual ads and image generation test
• Microsoft MAI-Image-2.6-Flash launch
• ChatGPT usage claims, claimed vs verified