ChatGPT can now make animated GIFs. OpenAI released Images 2.6 on 4 November in ChatGPT, Codex and its API, and the headline addition is motion: a prompt or a photo you upload can come back as a looping GIF. OpenAI says Images 2.6 is available globally on every ChatGPT subscription tier, on the web, desktop and mobile apps, and The Verge reports it is rolling out to all ChatGPT users. For marketers and social teams, it turns a job that used to need a designer and an animation tool into a single request.
What OpenAI released
OpenAI's announcement describes three changes. ChatGPT can create GIFs, which the company pitches for animating a brand mascot, turning a product shot into a loop, making reactions for social media or sketching animated storyboards before a video shoot. Image generation is 50% faster than before. Editing is more precise, with characters, products and brand details kept consistent across rounds of changes.
The Verge, describing an example OpenAI shared, reports that once a GIF is made you can adjust its playback speed, view it frame by frame, edit individual frames and share it. It also notes that ChatGPT could not natively generate a GIF in one step before this update. OpenAI's own post shows the feature through examples from partners such as OpenArt and Town rather than a full feature list, so the frame editing details come from The Verge's report and the image OpenAI supplied to it.
How it compares with what came before
OpenAI says people now create 4 billion images a week across ChatGPT, Codex and the API, which it describes as more than half a billion a day. The Verge points out that OpenAI reported 3 billion a week in September, so weekly volume grew by about a third in roughly two months. That growth helps explain why image features keep arriving quickly.
For developers, the API gets two models. GPT-Image-2.6 Flare is the fast option, and OpenAI says it can return an image in as little as six seconds when quality is set to low, with better layouts for posters and product images. GPT-Image-2.6 Sunburst is the higher quality option, with better consistency across edits at medium or high quality. The API also gains image markup, which ChatGPT already had, SVG export and a new complexity setting that controls how much planning the model does before drawing.
The prompt structure that makes a clean loop
The most useful detail is in OpenAI's sample prompts. Each one asks for a single animation sprite sheet, a grid of frames that the GIF plays in order, rather than describing a video. The samples specify exactly 6 frames in a 3 by 2 grid, read left to right and top to bottom, with identical panel sizes, a fixed camera angle and the same character, outfit and background in every frame. They describe the motion step by step and ask for the sixth frame to flow back into the first, which is what makes the loop seamless. They also ban captions, logos, borders and panel labels.
A short version you can adapt reads like this: make a seamless looping GIF of your mascot or product doing one action, use exactly 6 frames with the same character, camera angle and background, show the motion step by step, make frame 6 flow back into frame 1, and add no text or logos. Keeping the action to one simple movement gives the model less to keep consistent across frames.
What OpenAI has not said yet
The announcement does not state a maximum length, frame count, resolution or file size for GIFs, and it does not say whether free users face lower daily limits than paid users. OpenAI points readers to a system card for its safety evaluations. Until those limits are clearer, treat the feature as a fast way to produce short loops and test it on your own assets before promising anything to a client.
Brand teams should also check outputs closely. A six-frame loop is short, and small changes in a logo, a product shape or a face between frames are easy to miss at full speed. Stepping through frame by frame, which The Verge says ChatGPT now supports, is the quickest way to catch them.
Who should try it first
- Marketers who want a moving version of a mascot or product shot for a campaign or a landing page.
- Social media managers who make reaction GIFs and memes by hand today.
- Video and creative teams who want a rough animated storyboard before a shoot.
- Developers building creative tools, who can compare Flare and Sunburst on speed and consistency.
We think this is the first ChatGPT image update many marketing teams will use weekly, because short loops are a staple of social posts and were among the slowest assets to make by hand.
What to do this week
- Pick one existing asset, such as a product photo or a mascot, upload it and ask for a 6-frame looping GIF using the structure above.
- Step through the result frame by frame and fix any frame where the brand details drift.
- Try two playback speeds and keep the one that reads clearly on a phone.
- If you build on the API, run the same request through Flare and Sunburst and compare time and consistency.
If you are new to ChatGPT, the ChatGPT course in Learn starts from the basics, and our piece on the new ChatGPT app layout explains where the image tools now sit.