ChatGPT can now make GIFs. OpenAI released Images 2.6 on 4 November 2026, an upgrade to its image-generation system that turns a written prompt or an uploaded photo into a short animated loop — and lets you adjust the playback speed and edit individual frames before sharing it.
The update is rolling out to all ChatGPT and Codex users on the web, desktop and mobile, across every subscription tier including the free plan. OpenAI says GIF creation is being switched on gradually, so some accounts will see it a little later than others, but there is no separate paywall: existing account usage caps still apply.
How the GIFs work
The workflow is deliberately simple. Describe a motion in plain language — a cat dancing, a product mascot waving, a cartoon office worker reacting to a falling stock chart — or upload a photo or sketch, and the model returns a short looping animation. OpenAI’s demonstrations keep the character, camera angle and background steady across frames, which is the hard part of animation: a GIF where the subject morphs between frames looks broken, while one that holds steady reads as intentional motion.
Once a GIF exists, you can change its speed — reports of the rollout put the range at 0.5x to 2x — and revise or replace individual frames without regenerating the whole clip. That frame-level control is what separates this from a one-shot video generator: the result is meant to be edited like a document, not accepted or discarded whole.
Reference images: your characters, reused
The second big addition is reference images. You can save images of yourself, a pet, a character or a product, and reuse them across new prompts without uploading them again. For anyone making a recurring mascot, a comic character or a consistent product shot, this removes the most annoying part of AI image tools: re-explaining the same subject every time and getting a slightly different version back. Combined with the GIF tools, it points at a clear use case — repeatable short animations for social posts, brand accounts and product pages.
Faster generation, and two new API models
For developers, OpenAI is splitting its image API into two models. GPT-Image-2.6 Flare is tuned for speed and high-volume, everyday creation; GPT-Image-2.6 Sunburst is tuned for image quality and consistency across repeated edits, and adds SVG export. OpenAI says the new API models generate images 50% faster than their predecessors, and the release also brings ChatGPT’s image-markup tools to the API, so teams can point at an exact area of an image when requesting a change. Pricing for Flare and Sunburst was not included in the announcement.
The scale behind the release is striking. OpenAI says ChatGPT and its API now generate more than 4 billion images a week, up from 3 billion in September. Images 2.6 is the successor to Images 2.5, which arrived in September — a two-month upgrade cycle for a product used at that volume.
Why this matters — and what it is not
The GIF is the internet’s oldest native visual joke format, and until now ChatGPT could only make stills. Adding short, editable animation turns the chatbot into a reaction-image and micro-content machine: social teams, small businesses and ordinary users can now produce a branded loop in a chat window instead of opening an animation tool. OpenAI’s own example is telling: a startup called Town used the model to animate a virtual assistant character that appears while users wait for a reply, replacing a conventional loading spinner.
But OpenAI is careful to frame this as short, looping motion, not video. A GIF is a few seconds of repeated movement, not a narrated scene. Anyone expecting a video generator will be disappointed; anyone who wants a moving reaction image, a mascot wave or a storyboard sketch will find the tool fits the job. The gradual rollout also means the fairest verdict will come once free-tier users — the largest audience — have actually tried it, not from launch-day demos.
Related reading: our report on the other major AI release this week, Tencent’s Hy4, which adds vision and computer use, and our earlier explainer on Google’s Gemini 4 Argon. OpenAI’s own announcement is the primary source for this article: Images, upgraded — introducing ChatGPT Images 2.6.
