DeepMind strengthens Gemini 3 with expanded multimodal image and language capabilities

DeepMind’s Gemini 3 models, powered by Google’s Nano Banana Pro hardware, now support multimodal processing across text, images and code. Early tests show improved reasoning and visual generation quality, narrowing the gap with leading foundation models. Google plans to embed Gemini 3 across Search, Workspace and Cloud, positioning it as a key driver of enterprise productivity. The update highlights Google’s competitive focus in advanced multimodal ai systems.