Technology

Gemini models

Gemini models: Google DeepMind's multimodal AI family (Ultra, Pro, Flash, Nano), built for state-of-the-art reasoning across text, code, image, audio, and video data.

Gemini models are Google DeepMind's core multimodal AI foundation, succeeding PaLM 2: they natively process five data types (text, code, image, audio, video) for seamless understanding. The family includes specialized versions: Gemini Ultra (complex tasks), Gemini Pro (broad scaling), and Gemini Nano (efficient on-device use). The latest iteration, Gemini 3 Pro, delivers world-leading performance, specifically enhancing reasoning and agentic capabilities. This technology powers Google's generative AI chatbot, Gemini, and integrates directly into products like Search, Gmail, and Docs, enabling next-generation development.