OpenAI releases Images 2.6 with faster generation, GIFs, and finer editing controls

OpenAI has released Images 2.6 and says it is globally available in ChatGPT, Codex, and the API. This is a new image-model launch, not a minor control added to one existing editor: it combines faster generation, iterative editing, and animated output in OpenAI's image workflow.

OpenAI says Images 2.6 generates images 50% faster than before. API developers can choose GPT-Image-2.6 Flare or Sunburst: Flare is positioned for speed, with the company saying low-quality requests can complete in as little as about six seconds; Sunburst targets detail and consistency across editing rounds at medium or high quality. These speed and quality statements are OpenAI launch claims, so teams should measure latency and cost using their own prompts, resolutions, and concurrency.

The release also adds a GIF workflow. Prompts or sketches can become looping animations for mascots, product assets, social reactions, or pre-video storyboards. In the API, image markup lets an application identify the region to change, while SVG export can move a design into web, app, or print production. A new complexity parameter controls how much planning the model applies, and image quality controls visual detail and fidelity.

Creative teams should evaluate more than a single attractive image. Establish fixed prompts and acceptance samples for posters, product imagery, and branded characters, then compare generation time, detail retention across edits, manual rework, and total API cost. Confirm current data controls, content policy, and pricing before sending customer assets into production.

Related directory entry