ChatGPT Images 2.6 is OpenAI’s new image tool, released on Wednesday, November 4, 2026, and it is already switched on for every ChatGPT and Codex plan on the web, desktop and phone apps. The big change is motion. You can turn a written prompt or a photo into an animated GIF, then change its speed or fix one frame at a time. You can also save reference pictures of yourself, a pet or a character and reuse them later.
Key facts (as of November 2026)
- OpenAI says Images 2.6 went live worldwide on November 4, 2026, across all ChatGPT and Codex subscription tiers.
- Developers get a pair of API models, GPT-Image-2.6 Flare and GPT-Image-2.6 Sunburst, listed at $8 per million image input tokens and $30 per million image output tokens.
- OpenAI says people now create 4 billion images a week with its tools, up from about 3 billion in September.
What is ChatGPT Images 2.6?
It is the latest version of the picture generator built into ChatGPT, and it follows Images 2.5 from September. OpenAI says the update makes images 50% faster and keeps characters, products and brand details more stable when you ask for several rounds of changes.
Here is what is new for everyday users:
- GIFs: a prompt or an uploaded picture becomes a short looping animation. After that, you can adjust the playback speed, edit single frames and share the result.
- Saved references: keep photos of yourself, your pet or a recurring character, so you don’t upload them again for each new image.
- Faster output: the same request should come back in about half the time, according to OpenAI.
Until now, ChatGPT had no one-step way to make a GIF, The Next Web noted. The feature also lands a few months after OpenAI closed Sora, its separate AI video app, in March. So for now, short loops are the main way to get moving pictures out of ChatGPT.
Who can use ChatGPT Images 2.6, and what does it cost?
OpenAI says the update is available globally on all ChatGPT and Codex tiers, which covers users in the US, the UK and the EU. However, the announcement doesn’t list how many GIFs or images each plan allows per day. Limits may therefore differ between free and paid accounts, so check the image tool inside your own account.
For developers, the price is set per token, the small units that text and images are split into before a model processes them. The table below uses OpenAI’s API pricing page as of November 5, 2026.
| API model | Built for | Image input | Image output | Text input |
|---|---|---|---|---|
| GPT-Image-2.6 Flare | Speed; posters and product shots; as little as 6 seconds per image at low quality | $8.00 | $30.00 | $5.00 |
| GPT-Image-2.6 Sunburst | More detail and steadier results across edits at medium or high quality | $8.00 | $30.00 | $5.00 |
Prices are per million tokens. They match the list prices OpenAI shows for the older 2.5 models, so the upgrade doesn’t raise the rate card.
What changes for developers and designers?
Apart from GIFs, OpenAI is moving image markup into the API. That lets an app mark a precise spot on a picture and change only that area. Images can also be exported as SVG files, a vector format that stays sharp at any size on websites, in apps and in print.
There is also a new “complexity” setting. It controls how much planning the model does before it draws. OpenAI suggests low complexity for most quick edits and higher settings for crowded scenes with many rules. Early partners named by OpenAI include Figma Weave, Picsart, fal and OpenArt.
How does ChatGPT Images 2.6 label AI-made pictures?
OpenAI’s system card for the release says images carry C2PA metadata, an industry standard that records where a file came from. They also get an invisible SynthID watermark from Google DeepMind, applied across ChatGPT, Codex and the API. OpenAI adds that a public verification tool can check supported files for these signals.
The same document is candid about limits. In OpenAI’s own test of tricky GIF requests, its filters caught 77.7% of the animations labeled as breaking its rules. Motion-based violations “remain challenging to detect,” the company wrote. Separately, on 3,077 adversarial test prompts, the faster Flare model let through 2.14% unsafe images after moderation, against 1.49% for the older version. OpenAI said that difference was not statistically significant.
Why do ChatGPT Images 2.6 GIFs matter for news readers?
GIFs spread fast on social media, and a realistic loop can travel much further than its caption. Metadata can also be stripped when a file is screenshotted or re-uploaded. In short, a missing label proves nothing.
When a dramatic clip or picture appears during breaking news, look for the same scene from a named outlet before you share it. Our guide to spotting AI fakes during elections walks through the checks that also work for everyday news. For more coverage like this, see our AI news section.
Where can you follow AI and tech news live?
Tech launches often move markets, and business channels cover them first. The CNBC live stream guide and our Bloomberg TV guide show the official ways to watch in the US and Europe. For wider stories about AI and misinformation, the BBC News live stream and the CNN live stream are good options.
Coming soon: the NewsLive streaming app. We’re building one subscription for live international news on Android, Roku, Fire TV and desktop. The service is under development. See what the NewsLive app will offer and join the early-access list.
Frequently asked questions
Is ChatGPT Images 2.6 free?
OpenAI says it is live on all ChatGPT subscription tiers. It hasn’t published per-plan limits, so free users may get fewer images or GIFs than paying users.
Can ChatGPT Images 2.6 make videos?
No. It makes short animated GIFs and still images. OpenAI’s Sora video app shut down in March, according to The Next Web.
How can I tell if a picture was made with ChatGPT?
Files keep C2PA metadata and a SynthID watermark, according to OpenAI, and its verification tool can read them. Screenshots and edits can remove some signals, though, so also check the source.
