Google DeepMind has launched Gemini Robotics 2, a new AI model designed to control humanoid and other robotic systems. Building on its earlier Gemini Robotics technology, the new model enables whole-body control, more advanced object manipulation with robotic hands and grippers, and multi-robot collaboration.
The announcement comes as the United States is tightening restrictions on Chinese robotics. Earlier this week, the US FCC (Federal Communications Commission) banned new models of foreign-made humanoid and quadruped robots from entering US market over national security concerns. The move is widely seen as targeting China’s robotics industry.
According to Google DeepMind, its previous robotics models controlled a humanoid’s upper body for table-top tasks, while Gemini Robotics 2 expands to whole-body control, enabling movements such as walking, crouching, stretching and object manipulation. “For the first time, our model can now control entire humanoid robots, translating intent into intelligent whole-body control,” the company said.
The company said the model can control full humanoids, from feet to fingertips, and other bi-arm robots, while supporting dexterous manipulation using both hands and grippers.
Google DeepMind CEO Demis Hassabis also highlighted the launch in a post on X, saying Gemini Robotics 2 enables robots to “reason through every movement” to perform tasks that were not previously possible.
Gemini Robotics 2 is here, with our new suite of models, robots can now reason through every movement to manage tasks that weren’t possible before, like tying delicate knots – and even team up to solve complex workflows. Huge congrats to the robotics team on this great milestone! https://t.co/RajaoGIpF6
— Demis Hassabis (@demishassabis) July 30, 2026
Apart from Gemini Robotics 2, Google DeepMind also unveiled Gemini Robotics ER 2 and Gemini Robotics On-Device 2. Gemini Robotics ER 2 is a vision-language model (VLM) that enables robots to communicate with humans, understand the physical world and plan multi-step tasks lasting several minutes.
The company also introduced the ability for multiple robots to work together as a team. Gemini Robotics On-Device 2 is designed to run locally on robotic devices and can adapt to completely new robot embodiments with just a few hours of data.
The company said safety remains a key focus of its robotics research as robots become more capable of performing physical tasks. To improve safety, Google DeepMind introduced ASIMOV-Agentic, a benchmark for agentic safety orchestration and uncertainty resolution.
According to Google DeepMind, Gemini Robotics 2 is designed to navigate real-world environments and collaborate alongside humans, while its safety approach combines traditional physical safeguards with AI safety frameworks.
Also Read: China Makes AI Robots Core of Strategy in 15th Five-Year Plan: IFR
