OpenAI's Images 2.6 API adds SVG export, inline edits, and GIFs

Two API models — Flare for speed, Sunburst for fidelity — bring point-to-edit markup, a complexity dial, and animated GIFs; output is a reported $30 per 1M image tokens.

Nowline NOV 6 8:00 PM banner

Top AI stories from the last hour

Top AI stories from the last hour

Copy markdown

  • Two models: Flare for speed, Sunburst for fidelity

    The API now serves GPT-Image-2.6 Flare, which returns a low-quality draft in as little as ~6 seconds, and GPT-Image-2.6 Sunburst for higher detail and better subject consistency at medium or high quality. OpenAI says generation runs roughly 50% faster than 2.5 — fast enough to put image gen in an interactive loop.

  • Point at what to change

    You can now hand the model a marked-up image showing exactly where an edit should happen instead of describing it in prose, for tighter, more predictable edits when a pipeline iterates on a base image. SVG export also lands, so you can pull scalable vector output, not just raster.

  • A complexity dial and animated GIFs

    A new complexity parameter lets you trade detail against speed and cost per call, and both models can now generate animated GIFs — a genuine new modality to build on. Saved reference images keep a consistent style across a batch of generations.

  • What it costs

    Reported API pricing for both models is about $8 per 1M image-input tokens, $30 per 1M image-output tokens, and $5 per 1M text-input tokens — check OpenAI's pricing page before you budget a batch job. OpenAI says it is already serving roughly 4 billion images a week.