OpenAI launches ChatGPT Images 2.6 with GIFs, reference images and faster API models

OpenAI has rolled out ChatGPT Images 2.6, adding native GIF creation and saved reference images in ChatGPT, and saying two new API image models generate pictures about 50% faster. The company announced the upgrade on X on 4 November: “Images, upgraded—introducing ChatGPT Images 2.6. Now with GIFs and reference images. Plus, two new models are coming to the API with 50% faster image generation.”

The product post is titled Introducing Images 2.6 on OpenAI’s site. The Verge reported that ChatGPT can now create GIFs from a prompt or an uploaded photo, and that users can change animation speed, edit individual frames and share the result. Before this update, ChatGPT could not natively generate GIFs in a single step, according to that report.

What Images 2.6 adds in ChatGPT

According to OpenAI’s X post and The Verge’s write-up, the consumer-facing change centres on two features that were missing from earlier ChatGPT image tools: animated GIFs and reference images. The Verge describes an OpenAI example in which a user asks for a dancing-cat GIF and then adjusts playback speed or inspects frames one by one inside the chat.

Reference images let people save pictures to guide later generations, so a look or subject can be reused without re-uploading every time. OpenAI’s framing treats that as part of the same Images 2.6 release rather than a separate product. The Verge also notes OpenAI’s claim of faster image generation inside ChatGPT itself, alongside the API speed claim in the company’s X post.

The rollout is described as accompanying the new Images 2.6 model. Availability can still vary by plan and region as OpenAI stages updates; neither OpenAI’s short X announcement nor The Verge’s report claims every account receives every control at the same instant.

Flare and Sunburst in the API

OpenAI’s X post says two new models are coming to the API with 50% faster image generation. In OpenAI’s naming for this release, those models are GPT-Image-2.6 Flare and Sunburst. The “about 50%” figure is OpenAI’s own comparison; the company has not published an independent third-party benchmark in the sources reviewed here, so treat the percentage as a vendor claim.

Developers who already call OpenAI’s image endpoints are the intended audience for Flare and Sunburst. Codex is included in OpenAI’s broader Images 2.6 messaging alongside ChatGPT and the API, according to the product framing in the story pick and OpenAI’s announcement package. Exact model IDs, rate limits and pricing should be checked on OpenAI’s developer documentation when the models are listed there.

Why the GIF step matters

Animated output matters because ChatGPT was already used for static marketing art, memes and product mock-ups. Native GIFs keep that workflow inside one chat instead of exporting stills to a separate animation tool. Frame-level edits and speed controls, as described by The Verge, also push the feature beyond a one-shot novelty.

The release does not by itself settle questions about copyrighted style imitation, deepfake misuse or watermarking. Those topics sit outside the product announcement and are not adjudicated in OpenAI’s short X post.

Images 2.6 is therefore a concrete feature drop: GIFs and reference images in ChatGPT, plus Flare and Sunburst for API users OpenAI says can generate images roughly half as slowly as before. How widely those gains show up in day-to-day latency will depend on load, prompt complexity and which model each product actually routes to.

This report is not legal, financial, or health advice. See the disclaimer.