The short answer
● Launched — Images 2.6, 4 November
● Two API models — Flare for speed, Sunburst for quality
● Speed — 50% faster, as little as six seconds at low quality
● New outputs — GIF creation and SVG export
● Price — $30 per million image output tokens
OpenAI introduced Images 2.6 on 4 November across ChatGPT, Codex and the API. It splits into two API models: GPT-Image-2.6 Flare, tuned for speed and quick iteration, and GPT-Image-2.6 Sunburst, tuned for output quality and consistency across repeated edits at medium or high quality.
OpenAI claims images generate 50% faster than before, with Flare producing one in as little as six seconds at low quality. That is a best case rather than a guaranteed response time. Pricing is published as $30 per million image output tokens rather than a flat per-image rate, so the real cost moves with output size, quality and any text or image inputs. The new complexity parameter carries no separate fee.
The two genuinely new output formats are GIF creation and SVG export. SVG has real restrictions worth knowing before you build on it: it requires n=1 and supports neither streaming, partial images, output compression nor the prompt rewriter, and the image-edit endpoint will not return it at all.