OpenAI released its ChatGPT Images 2.5 model on September 8. It reduces generation latency by up to 50% compared to the previous version and supports over 3 billion image generations per week.
The company launched two API variants: GPT-Image-2.5 Flare for speed and quality, and GPT-Image-2.5 Sunburst for precision. Both are available in ChatGPT, ChatGPT Work, and Codex.
Google launched its competing model, Nano Banana 2, based on Gemini 3.1 Flash Image, on February 26. Independent tests show Google retains an advantage in structured prompt precision.
OpenAI outperforms Google in creative tasks, including natural blending, visual depth, and multi-turn editing. The new model features in-chat sketching, which allows users to convert rough outlines into polished images.
Both companies now embed C2PA metadata in generated images and support reference images with multiple aspect ratios.
