OpenAI's Images 2.6 Makes GIFs, SVG Exports and 50% Faster Generation

What Shipped: Images 2.6 Across ChatGPT, Codex and the API

OpenAI rolled out Images 2.6 on November 4, 2026, and it's live globally today across ChatGPT, Codex and the API — every subscription tier, on desktop, mobile and web. The scale behind the update is the story as much as the features: people now create roughly 4 billion images per week across OpenAI's surfaces, more than half a billion every single day. Image generation has quietly become one of the most-used capabilities in consumer software, and OpenAI is iterating on it like a flagship product.

The headline changes are easy to summarize: generation is 50% faster than before, images can now be animated as GIFs natively, and developers get a split into two API models — GPT-Image-2.6 Flare for speed and GPT-Image-2.6 Sunburst for quality and consistency. Underneath that sit quieter but arguably more important upgrades: better character, product and brand consistency across editing rounds, image markup in the API, SVG vector exports, and a new complexity parameter that controls how much planning the model does before drawing.

GIFs Are the Headline Feature

For the first time, ChatGPT can turn a prompt — or an existing image — into an animated GIF. You can adjust the animation speed, edit individual frames, and export a loop that's ready to drop into a chat, a README or a social post. OpenAI's framing is deliberately practical: marketers animating a brand mascot for a campaign, creators making reaction memes, and studios using animated storyboards to explore character movement before committing to full video production.

The most striking demo is the sketch-to-animation pipeline. OpenAI showed a pen-on-lined-paper owl drawing transformed into an expressive animated character — the kind of thing that used to require a dedicated animation workflow now happens from a photo of a napkin sketch. Startup Town is already using the API to animate "Monty," a character that appears as a live typing indicator in its AI work assistant.

It's worth being precise about what this is: GIF generation, not video generation. You get short loops and sprite-sheet-style sequences, not the multi-second cinematic clips that dedicated video models produce. But for the enormous middle ground of use cases — reactions, mascot loops, micro-animations in products — a GIF from a single prompt removes the need to touch a video tool at all. As OpenArt CEO Coco Mao put it in the launch post: "GIFs are the native language of social media, and GPT-Image-2.6 lets every OpenArt creator speak it fluently from a single prompt."

Flare vs. Sunburst: The New Two-Tier API

Developers calling the image API now choose between two models with distinct personalities:

• GPT-Image-2.6 Flare is the speed tier. With quality set to low, Flare generates an image in as little as six seconds. OpenAI also says Flare improves layout quality for posters, product imagery and other professional visuals — so "fast" no longer means "slapped together."
• GPT-Image-2.6 Sunburst is the quality tier. At medium or high quality settings it delivers higher-fidelity generation and, critically, greater consistency across edits — the same character or product stays recognizably the same as you iterate, which has always been the weak point of conversational image editing.

The two-tier split mirrors what has worked well elsewhere in the model market: a cheap, fast option for high-volume pipeline work and a premium option for hero assets. If you're generating a hundred thumbnails for A/B testing, Flare is your tool; if you're refining a key brand illustration across a dozen edit rounds, Sunburst pays for itself in consistency.

Pro Editing Controls: Markup, SVG and Complexity

Three developer-facing additions matter more than they might first appear:

• Image markup in the API. The markup feature — already familiar from ChatGPT, where you circle or point at the region you want changed — is now programmable. Teams can pinpoint exactly where an edit applies instead of describing locations in prose and hoping the model finds them.
• SVG export. Vector output means generated designs can scale across websites, apps and print without rasterization artifacts. For logo-adjacent and UI-adjacent work, this bridges a real gap between AI generators and professional design tools.
• The complexity parameter. Developers can now dial how much planning the model puts into a request — low complexity for quick edits, medium or high for scenes with many elements or constraints. OpenAI's guidance: pair low complexity with the image quality your workflow needs for most editing tasks, and reserve high complexity for dense, rule-heavy scenes.

Figma Weave is already integrating the model, with Creative Director Itay Schiff highlighting "speed and creative control together" inside reusable team workflows, and Picsart reports faster generations with better character consistency for its creators.

How It Stacks Up Against the Best AI Image Tools

Where does this leave the rest of the AI image landscape? The honest answer is that Images 2.6 attacks the convenience and iteration layers, not necessarily the ceiling of quality.

Midjourney still leads on pure photographic richness and stylized art direction. Ideogram remains the pick when the image is mostly type — logos, posters, packaging. Leonardo AI bundles generation, editing and upscaling with its own motion features and a daily free allowance. And Flux keeps the open-weights crown for anyone who wants generation with no vendor dependence. What Images 2.6 offers that none of them match is frictionlessness: it's already in the chat window a billion people use, it edits as naturally as it generates, and now it animates and exports vectors from the same conversation.

Pricing and Availability

Images 2.6 is available globally today in ChatGPT and Codex across all subscription tiers — including the free tier — on desktop, mobile and web. GPT-Image-2.6 Sunburst and GPT-Image-2.6 Flare are available in the API, with per-image pricing detailed on OpenAI's developer pricing page. OpenAI has also published a system card covering the safety evaluations and safeguards for the new model, including the motion-related safeguards that come with animation capability.

What This Means for Creators

The gap between "AI image generator" and "AI design tool" just narrowed again. Three practical takeaways:

Frequently Asked Questions