OpenAI Releases Images 2.6 With GIFs and SVG Export
OpenAI released Images 2.6 on November 4, adding animated GIFs alongside still-image generation. The update also brings SVG output and more editing controls to the API.
The company reports 50% faster image generation. Flare, the faster of the two API models, produces images in as little as six seconds at low quality, OpenAI says. Sunburst targets detailed compositions and more consistent edits at medium or high quality.
OpenAI says the updated models better preserve characters, products, and visual details through successive edits. Sketches and annotations give the models more direction. Image markup, already available in ChatGPT, now reaches the API, letting developers indicate the part of an image they want changed.
The API documentation specifies GIFs with 2 to 64 frames and configurable timing. The default uses nine frames at 200 milliseconds each, producing a 1.8-second animation. GIF output works through the image-generation endpoint, rather than the image-edit endpoint.
SVG output follows the same endpoint restriction. The API generates a reference image, then converts the result into static vector content with editable paths, shapes, and colors. SVG output excludes scripts, external resources, and embedded raster images.
A separate complexity setting controls how much planning happens before rendering. Developers choose low, medium, or high, with high as the default. Quality controls rendering fidelity. Complexity suits a different decision: a focused background change needs less planning than a dense layout with several objects. OpenAI recommends low complexity for most edits and warns higher settings sometimes introduce unwanted changes or increase latency.
Both GPT-Image-2.6 Flare and GPT-Image-2.6 Sunburst accept text and image inputs for generation and editing. OpenAI positions Flare for interactive workflows and Sunburst for detailed visual work.
OpenAI lists the same standard token prices for both: $5 per million text input tokens, $8 per million image input tokens, and $30 per million image output tokens. Cached inputs receive lower rates. These are token charges, rather than fixed prices per image. Changing complexity carries no separate fee.
Images 2.6 launched globally in ChatGPT and Codex across all subscription tiers, on desktop, mobile, and web. Both API models have been available since the release date.