OpenAI Images 2.5 launches with Adobe Firefly as first named integration

OpenAI released ChatGPT Images 2.5 on September 8, 2026, with up to 50% lower latency than Images 2.0 and improvements to detail, lighting, texture, multi-turn instruction-following, and subject preservation in reference photos. Two API model variants ship with the release: GPT-Image-2.5 Flare, the default for fast everyday generation, and GPT-Image-2.5 Sunburst, a slower, higher-control option for production creative work. Adobe is a launch customer and has integrated the model into Firefly. The release rolls out across all ChatGPT tiers on desktop, mobile, and web. A Sketch feature lets users draw directly in ChatGPT as a visual reference for the model to render, and a prompt-sharing feature lets users distribute prompts alongside generated images for remixing. OpenAI reports more than 3 billion images are generated weekly across ChatGPT Images and the GPT-Image API, though the figure is self-reported. Pricing, architecture details, and numerical benchmarks comparing Flare and Sunburst have not been officially confirmed.