Images 2.6 in ChatGPT: GIFs, SVG Export and Faster Edits

OpenAI has released Images 2.6, an update that adds animated GIF creation to ChatGPT and shortens the wait between an idea and a usable picture. The headline figure is speed: OpenAI says the model generates images 50% faster than before. For researchers who follow generative image systems, the more interesting shift is control, with consistency across editing rounds and new settings for how much planning the model does before it draws.

• OpenAI says Images 2.6 generates images 50% faster and supports GIFs for looping animation
• The API offers two models: GPT-Image-2.6 Flare, as fast as six seconds at low quality, and GPT-Image-2.6 Sunburst for higher quality
• Developers also get image markup, SVG exports and a new complexity parameter
• OpenAI reports people now create 4 billion images per week across its products

What Is New in Images 2.6?

The most visible addition is motion. Images 2.6 can animate a brand mascot, turn a product shot into a loop or build a rough animated storyboard ahead of video production. OpenAI says even a simple sketch can become an expressive animated character. It builds on the sketch and latency work in Images 2.5, which arrived in September.

OpenAI also emphasizes consistency, meaning characters, products and brand details should hold steady as a user revises an image. That has been a persistent weakness of image generators, since each edit could quietly change a logo or a face. The company attributes this to Sunburst in particular at medium and high quality settings, while it credits Flare with better layout for posters and product imagery. These are vendor claims, and Decrypt noted before the full announcement that it could not independently test the model, so independent comparisons will be worth watching.

How Do GPT-Image-2.6 Flare and Sunburst Differ?

The two API models represent a familiar trade-off between latency and fidelity:

• Flare suits rapid iteration, and OpenAI says it can return an image in as little as six seconds at low quality
• Sunburst targets detailed work, with stronger consistency across edits at medium or high quality

A second dial is arguably the more novel one. The new complexity parameter sets how much planning the model does for a request, separately from image quality, which governs visual detail and fidelity. OpenAI recommends low complexity for most edits, and medium or high for scenes with many elements or constraints. In effect, developers can spend compute on composition and on rendering independently.

Image markup, already in ChatGPT, also reaches the API so teams can point to the exact region they want changed. SVG exports make output easier to reuse across websites, apps and print.

Who Can Use ChatGPT Images 2.6?

OpenAI says Images 2.6 is available globally in ChatGPT and Codex across all subscription tiers on desktop, mobile and web, with Flare and Sunburst in the API. Early partners quoted in the announcement include Picsart, OpenArt and Figma Weave, and the safety evaluations are detailed in a system card.

The scale figure gives useful context here. At more than half a billion images a day, small reductions in latency add up quickly for both users and infrastructure. You can follow more model releases in our AI coverage.

Sources: OpenAI — November 4, 2026; Decrypt — November 4, 2026.