OpenAI brings GPT-Image-2.6 Flare and Sunburst to the API and Codex

OpenAI released Images 2.6 on November 4 in ChatGPT, Codex and the API, available globally in ChatGPT and Codex on all subscription tiers. In the API it comes as two models, GPT-Image-2.6 Flare and Sunburst. OpenAI says both generate images 50% faster than before. Flare, at low quality, can produce an image in as little as six seconds and improves layout for posters and product imagery. At medium or high quality, Sunburst gives higher-quality images and keeps details more consistent from one edit to the next. ChatGPT Images 2.6 adds GIF output and reference images. OpenAI says people now create 4 billion images per week across ChatGPT, Codex and the API.

New for developers, per OpenAI:

• Marked-up edits: image markup, already in ChatGPT, comes to the API, so an image with markup can show where an edit should go.
• Complexity parameter: sets how much planning the model does for a request, separately from image quality, which controls detail and fidelity. OpenAI recommends low complexity for quick edits and medium or high for scenes with many elements or constraints.
• Animated GIFs: a still character, sprite or mascot can become a looping GIF from a description of the motion.
• SVG export and, in Codex, editing directly in the overlay instead of a side panel.

OpenAI quotes early-access customers including OpenArt, Figma Weave and Picsart, and published a system card for the release.

Arena (DesignArena) says GPT Image 2.6 Sunburst ranks #1 on its Image Arena, ahead of Image 2.5 Sunburst, and that the Flare and Sunburst models set a new preference-versus-speed frontier for generation and editing. It reports Flare as 2.6× faster than its predecessor and Sunburst as 1.15× faster. Those are Arena's own figures.

In a separate post, Arena's main leaderboards place Sunburst (Medium) #2 in Text-to-Image with 1409 points and Flare (Medium) #4 with 1384, and Sunburst #3 in Image Edit with 1476 points and Flare #5 with 1462. Arena says Sunburst is #1 in the Cartoon, Anime & Fantasy category and that OpenAI holds the top five Text-to-Image spots. It repeats OpenAI's claim of more than 2x faster generation than GPT Image 2.5 at similar quality, with latency as low as 5 seconds.

On November 6, Artificial Analysis said GPT Image 2.6 takes #1 and #2 on both its AA-Image-T2I v2.1 (text-to-image) and AA-Image-Editing v2.1 leaderboards. It says Flare is almost twice as fast as GPT Image 2.5 Flare with slightly better quality, and that Sunburst takes the top spot with longer generation times. Those are Artificial Analysis's own results.