OpenAI is rolling out animated GIF creation in ChatGPT, turning the chatbot’s image tool from a stills maker into a short-animation maker. Users can ask for a GIF from a written prompt or an uploaded photo, then adjust how fast it plays, inspect the frames behind it, edit individual frames and share the result. The feature arrives with the new Images 2.6 model and is reaching all ChatGPT and Codex users on the web, desktop and mobile apps.
A new format, not just a new filter
Until now, ChatGPT could not produce a GIF in a single step. Users who wanted a looping animation had to stitch still images together or turn to a separate tool. Images 2.6 closes that gap by generating the sequence itself, so the request, the animation and the edits all live in one conversation.
The example OpenAI showed off was deliberately ordinary: a user asks for a GIF of a cat dancing, and ChatGPT returns a short loop of the animal mid-dance. The demo matters less for the cat than for what it proves about consistency, since the character has to stay recognisable from frame to frame or the loop falls apart.
Once the GIF exists, the user can change its playback speed, look at the frames one by one and revise a single frame without rebuilding the whole animation. That frame-level control is the more useful half of the update, because a loop that cannot be fixed is usually a loop that gets discarded and remade from scratch.
What Images 2.6 adds for developers
For developers, OpenAI is releasing two API models alongside the ChatGPT rollout. GPT-Image-2.6 Flare is positioned for faster generation, while GPT-Image-2.6 Sunburst is positioned for higher-quality output. The split mirrors the Images 2.5 lineup, which already separated throughput from finer creative control, so existing developers face a familiar choice between speed and quality.
OpenAI says users now generate more than 4 billion images a week across ChatGPT and its API, up from more than 3 billion a week when Images 2.5 was announced on September 8. That figure is OpenAI’s own tally, not an independently checked count, and it covers all image activity rather than GIFs alone.
Images 2.6 is also meant to make image generation faster, according to the reports. Speed matters at this volume, because a tool used billions of times a week lives or dies on waiting time. A faster model also lowers the cost of experimenting, which is what a new format like GIFs needs while users learn what it is good for.
Why It Matters
GIFs sit in a gap that still images and video do not fill. A still image cannot show motion, and a generated video is heavier to make, longer to watch and harder to drop into a chat. A short loop is light, instantly readable and built for sharing, which is why the format has survived decades of newer media.
Giving ChatGPT native GIFs therefore extends its reach into everyday communication, not just creative work. Reactions, explanations and jokes are the use cases OpenAI’s own demo points to, and they are also the uses most likely to push weekly image totals higher still.
Our read
Our read is that the GIF move is pragmatic. OpenAI is not trying to outdo video generators at their own game; it is claiming the lightweight end of motion, where speed and shareability beat length and polish. The frame-editing controls support that reading, since they treat a GIF as something to fix and reuse rather than a finished clip.
The risk for OpenAI is the opposite of the opportunity. If GIFs flood chats with low-effort loops, the format’s charm wears thin, and the feature reads as a novelty rather than a tool. That is the fate of many chatbot additions that launch with a memorable demo and no lasting job to do.
The character-consistency problem cuts the other way. Once users trust that a subject stays stable across frames, they will ask for longer, more complex motion, and a short loop may start to feel like a constraint rather than a convenience. That pressure points back toward full video tools, which is where the longer stories will go.
Background
Images 2.5, announced on September 8, improved fine detail, lighting and texture rendering and consistency across successive edits, while cutting generation latency by as much as half against Images 2.0. It also added sketch guides, templates and comment-based revisions, and Images 2.6 builds on that base with native animation.
Source: Unite.AI’s report and The Decoder’s report, both covering The Verge’s November 4 report.
0 Comments