GPT Image 2.6 is OpenAI's next image generation model, announced on November 4, 2026. Like GPT Image 2.5, it ships as two variants: Sunburst, which spends more compute on complex scenes and precise edits, and Flare, which prioritizes speed. OpenAI reports the new generation runs roughly 50% faster, and native GIF output arrived alongside it in ChatGPT.
Both variants are live in the generator above. Every sample on this page - including the two animated GIFs - was generated right here with GPT Image 2.6 Sunburst through the site's own pipeline, no mockups. Draft on Flare when you want speed, finish on Sunburst when you want maximum fidelity.
If you draw characters, illustrate stories, or storyboard animations, the two generations share the same prompt structure, so everything you learned prompting GPT Image 2.5 carries straight over to 2.6.
OpenAI, announced November 4, 2026
Sunburst (quality-first) and Flare (fast), same split as 2.5
Live - both variants are in the generator above
The stained-glass whale comes from OpenAI's official announcement; the other three examples are generated right here with GPT Image 2.6 Sunburst through our own generator.
The announcement's stained-glass whale shows where the speed still leaves room for detail: refraction, translucency, and layered light across a whole scene. Surfaces like these are what cartoon backgrounds and fantasy illustrations keep asking for.
A four-panel pencil storyboard of a small robot learning to fly, generated from one prompt: loose graphite lines, numbered panels, and handwritten captions all read clean. The model that renders your polished scenes can also sit down and sketch.
We asked for one sparrow on the nearest lantern of a crowded ramen-shop scene, and everything else held: the cat chef, the neon signs, the menu board, the rain-slicked street. Targeted edits like this are what character sheets and recurring scenes are built on.
Our balloon-letter BOUNCE poster makes the typography point, and GIF mode turns prompts into looping animations - the animated examples on this page are real outputs. SVG exports are named in the announcement for designs that scale across web, apps, and print; we switch them on here as soon as the provider's endpoint delivers them.
Real outputs from this site's generator: a looping GIF sticker, vector-style logo art, and a photo turned into a cartoon character.
Prompt
a small round penguin chef flipping a pancake in a tiny skillet, the pancake tossing up and falling back, joyful loop
Set the output format to GIF and a prompt becomes a looping animation - reaction stickers, emotes, and playful loops straight out of the model.
Prompt
a flat vector style mascot logo of a friendly cartoon fox head inside a circular badge, bold clean shapes, minimal flat colors, sticker design
Flat, bold shapes for mascots, badges, and sticker art. SVG export is the announced next step - today the same prompts already give you print-ready flat artwork.
Hand the model a photo and it redraws the subject as an animated-movie style character while keeping the pose, the expression, and the setting recognizable.
What the release changes for character designers, illustrators, and animators - all of it usable in the generator above.
Try GPT Image 2.6 nowOpenAI reports roughly 50% faster generation. Cartoon work means rerunning a prompt many times to nail a pose or expression, so every saved second compounds across a session.
Sunburst and Flare keep the quality-versus-speed split from 2.5: draft compositions on Flare, finish the winner on Sunburst. Your prompt library moves between them unchanged.
GIF output is live through the API, not just ChatGPT: this page's animations are real prompts turned into looping stickers. Short animated moments for reactions and simple loops, not just stills.
Both variants run on this page today: pick Sunburst or Flare in the model picker, switch the output format to GIF when you want motion, and layer the site's cartoon styles onto your prompts.
OpenAI's next image generation and editing model family, announced on November 4, 2026. It succeeds GPT Image 2.5 and ships as two variants: Sunburst for maximum fidelity and Flare for lower latency, the same split as the previous generation.
Toonderly is an independent third-party tool built on publicly available model APIs. "GPT Image", "OpenAI", and related marks are trademarks of OpenAI. The stained-glass whale feature image comes from OpenAI's official Images 2.6 announcement materials and remains the property of its owners; every other sample image and animation on this page was generated on this site with GPT Image 2.6. This page is not affiliated with, sponsored by, or endorsed by OpenAI.
