On July 30, 2026, Google DeepMind made a significant announcement—the introduction of three cutting-edge AI models tailored for robots: Gemini Robotics 2, Gemini Robotics ER 2, and On-Device 2. Among these, Gemini Robotics 2 stands out by empowering humanoid robots to synchronize their full-body movements. This enables them to perform a range of actions, including walking, squatting, and object manipulation, all while undertaking complex reasoning tasks and orchestrating the entire humanoid robot's operations. Gemini Robotics ER 2, a substantial upgrade from its predecessor Gemini Robotics ER 1.6, is hailed as a sophisticated brain for robots. It facilitates seamless, natural interactions between robots and humans, deepens their understanding of the physical world, and enables them to plan and execute multi-step tasks with precision. On the other hand, On-Device 2 operates locally on robots, eliminating the need for an internet connection. This model showcases remarkable versatility and adaptability to various tasks. Collectively, these three models can function in tandem or independently, driving the broader adoption of robots across diverse scenarios.