The GIF is the internet's oldest visual joke, and OpenAI just industrialized it. The company announced Images 2.6 on November 4, an update that lets ChatGPT turn a text prompt or an uploaded photo into a looping animated GIF — no converter app, no timeline editor, no export step. Describe the animation, or hand it a picture and ask it to move. You can then adjust the playback speed, inspect and edit individual frames, and share the result on a loop.
The headline feature is the GIF maker, but the more consequential change is what sits behind it. Images 2.6 also introduces saved reference images: upload a photo of yourself, a pet, or a character once, and ChatGPT keeps it on file to reuse across new creations. Combined with OpenAI's claim that the new model holds characters, products, and brand details consistent across edits — the thing every AI image tool has historically failed at the moment you ask for a second version — this is a shift from one-off generations to reusable visual assets. A brand mascot that survives ten revisions is a different product than a mascot that changes faces every time you tweak the background.
Speed is the other half of the pitch. OpenAI says Images 2.6 generates images 50% faster than its predecessor, and the scale explains why that number matters: people now create around 4 billion images a week across ChatGPT, Codex, and the API, up from 3 billion in September. At that volume, a faster model isn't a convenience feature — it's the difference between the product keeping up with its own demand and falling behind it.
Developers get the most granular version of the update. Two new API models split the workload: GPT-Image-2.6 Flare, built for speed and able to produce an image in as little as six seconds at low quality, and GPT-Image-2.6 Sunburst, aimed at detail and consistency across edits at medium or high quality. The API also gains markup-based editing, SVG export, and a complexity setting that controls how much planning the model does before generating.
There are caveats worth stating plainly. The GIF tool is rolling out gradually, so not every account has it yet — some early testers reported it missing entirely. And a convincing animated image of a real person spreads faster than a still one ever did, which is exactly the kind of thing OpenAI's safety evaluations for the release will be judged on. The company says Images 2.6 is available globally across all ChatGPT and Codex tiers on web, desktop, and mobile, with the new API models live now.
Watch what happens to the feed. Reaction GIFs and memes have always been handmade from existing footage; now the raw material is a sentence. The question isn't whether people will make GIFs — 4 billion images a week already answered that — it's whether the format survives being generated on demand, or whether the charm was always in the hunt.
— via Digital Trends