Gemini Robotics 2 brings whole body intelligence to robots

Google DeepMind has introduced Gemini Robotics 2, a set of AI models designed to give robots the ability to learn and adapt rather than simply follow pre-programmed instructions. The system enables robots to perform complex tasks requiring whole-body coordination, fine hand dexterity, and multi-step reasoning, while also allowing different robots to work together on shared problems. This matters because most existing robots are limited to narrow, repetitive tasks and cannot easily transfer skills between different robot bodies, whereas these new models can adapt to new robot types in just a few hours. The technology combines three components: a vision-language-action model for motor control, an embodied reasoning model for planning and communication, and an on-device model optimized to run locally without internet connectivity.