OpenAI released Images 2.6 on 4 November 2026, bringing GIF creation to ChatGPT and adding image-generation and editing controls for developers using its API. The company says the update also produces images 50% faster than before, while offering two API models for work that calls for different balances of speed and visual quality.
Key points
- ChatGPT can make GIFs from prompts or uploaded photos, with controls for playback speed and individual frames.
- OpenAI says Images 2.6 generates images 50% faster than before.
- The API offers Flare for quicker generation and Sunburst for higher-quality output.
Images 2.6 brings editable GIFs to ChatGPT
ChatGPT can turn either a written request or an uploaded photo into an animated GIF, The Verge reported. Once it has made the animation, users can change its playback speed, inspect it frame by frame, edit individual frames and share it. The publication reported that ChatGPT previously could not make a GIF natively in a single step.
A GIF creates movement by displaying still pictures in sequence, much like a flipbook. In examples published by OpenAI and created by OpenArt, requests for an animated character specify six successive frames arranged in a grid. They ask for the character, setting and camera position to remain consistent while the expression or action changes from one frame to the next, with the final frame leading back towards the first for a loop. That continuity matters when a sequence must read as one moving subject rather than a collection of separate pictures.
Making an animated reaction for a message could begin with a written description or an uploaded photo, then continue with a change to the playback speed or one frame before sharing. Inspecting the frames individually would make it possible to focus that change on a particular moment in the sequence.
OpenAI describes uses ranging from reactions and social-media posts to animated characters and product visuals. Its examples include a hand-drawn owl sketch turned into an animation and character sequences created by OpenArt. These are demonstrations of the kinds of output the company chose to publish, rather than measurements of how consistently the model handles other subjects or longer sequences.
Images 2.6 is rolling out to all ChatGPT and Codex users on the web and in desktop and mobile apps, The Verge reported. The GIF controls extend the editing path inside ChatGPT: a request can produce the animation, after which its timing or a particular frame can be revised there.
Flare and Sunburst divide API work
For developers, OpenAI offers GPT-Image-2.6 Flare and GPT-Image-2.6 Sunburst through the API. The company says Flare can produce an image in as little as six seconds when quality is set to low. It positions Sunburst for higher-quality generation and greater consistency between edits at medium or high quality settings. OpenAI also says Flare improves layouts for posters and product imagery.
The six-second figure is a claim about Flare at a particular quality setting, not a stated time for every image or for Sunburst. OpenAI separately says Images 2.6 generates images 50% faster than before. A latency comparison accompanying its announcement depicts the generation of a travel advertisement, with the timers shown at eight-times playback speed.
The two API options give an application a choice between a model OpenAI presents as quicker at low quality and one it presents as more detailed at higher settings. That distinction matters when images are generated as part of a larger process: waiting for a draft and preparing a finished visual place different demands on the same generation step. The company also says developers can adjust how much planning the model does for a request through a complexity parameter, with lower settings suited to quick edits.
Image markup, which OpenAI says was already available in ChatGPT, is coming to the API so a change can be directed to a particular part of an image. The principle is familiar from marking a spot on a printout rather than asking for the whole page to be redone. OpenAI says the API will also support SVG exports, a format for designs that can be used across websites, apps and print.
OpenAI reports 4 billion images a week
OpenAI says people create 4 billion images per week across ChatGPT, Codex and the API. In September, the company reported 3 billion images per week across ChatGPT and its API, The Verge reported. Those are company figures for image creation across its products, rather than a count of GIFs made with Images 2.6.
OpenAI also describes an API use beyond making a standalone image: Town, a startup building an AI work assistant, used Images 2.6 to explore animated characters for interactions at work. In the example OpenAI published, Town’s animated character Monty appears above a message box as a typing indicator.
