OpenAI launches Images 2.6 with GIFs and image reference

Open original

OpenAI has rolled out Images 2.6 across ChatGPT, Codex, and the API, adding GIF creation and better consistency across edits. Developers get two new models, Flare for speed and Sunburst for detail.

  • The update adds GIF creation, turning characters, products, and storyboards into short animated loops, and aims to keep characters and brand details consistent across successive edits.
  • OpenAI says images now generate 50% faster, and usage has reached 4 billion images per week across its products.
  • In the API, Flare can produce a low-quality image in about six seconds, while Sunburst targets more detail and steadier edits. A new complexity parameter and SVG export round it out.

OpenAI is rolling out Images 2.6 to everyone using ChatGPT and Codex, across all subscription tiers and on desktop, mobile, and web. The bigger news for developers is that two new API models, GPT-Image-2.6 Flare and Sunburst, are now available. OpenAI says the release generates images 50% faster than before, and it points to usage of 4 billion images per week across its products as the scale it's now building for. The headline feature is GIF creation. Users can turn characters, mascots, product shots, sketches, reactions, and storyboards into short animated loops. The update also tries to keep characters, products, and brand details consistent across repeated edits, which has been a persistent weak spot for image generators. Anyone who has tried to iterate on a logo or mascot over several rounds knows how quickly details drift, so that consistency claim is the part worth testing in practice. The two API models are split by purpose. Flare is built for speed and clean layouts like posters and product imagery, and at low quality it can produce an image in as little as six seconds. Sunburst targets more detail and more consistent edits at medium or high quality. The API also gets ChatGPT's image markup, so teams can point at exact regions to change, along with SVG export for web, app, and print work. A new complexity parameter sets how much planning the model does before rendering. OpenAI suggests low for most quick edits and saving medium or high for scenes with many elements or tight constraints, which is a sensible rule of thumb, since heavier planning costs time you don't need for simple changes.

Ask about this article