Gemini Robotics

Google DeepMind announced Gemini Robotics 2, its most advanced vision-language-action model that converts vision and language input into motor control, enabling robots to autonomously perform complex tasks with whole-body intelligence and advanced dexterity. The model can be adapted to any bi-arm robot in hours, supports multi-robot collaboration, and pairs deep spatial reasoning with long-horizon planning for unfamiliar situations. Google DeepMind has taken a comprehensive approach to safety, including collaborations with experts and its Responsibility and Safety Council.

Gemini Robotics 2 Our most advanced vision-language-action model VLA that converts vision and language input into motor control, enabling a robot to take action The intelligence layer to power any kind of robot Our most advanced VLA model: capable of intelligently controlling any type of robot, from bi-arms to full humanoids Our embodied reasoning model: capable of real-world understanding and complex, multi-step planning Our most efficient VLA model, optimized to run locally on robotic devices Our models enable robots of every shape to think, act, and interact with the world around them. With delicate precision and full-body mastery, they autonomously solve a range of complex tasks – using their intelligence to figure out new situations on the fly. Our vision-language-action VLA model and embodied reasoning ER model work together to interact with the physical world. Each has a specialist role, but they operate as one powerful and versatile system. Our most advanced vision-language-action model VLA that converts vision and language input into motor control, enabling a robot to take action Gemini Robotics 2 is the next step on our journey towards general, useful robotics. It brings whole-body intelligence to humanoids, enables advanced dexterity, and can coordinate multiple robots to work together in shared spaces. Most robots are trained to do one specific task over and over. But Gemini Robotics 2 can complete a variety of tasks – even if it hasn’t been trained on them before. It’s able to adapt to new and unfamiliar situations on the fly. Can be adapted to any bi-arm robot in just a few hours, scaling its intelligence from arms to complex humanoid bodies. Enabling a new level of dexterity that enables robots to complete delicate actions requiring finesse, like screwing in a light bulb and tying knots. Controls entire humanoid bodies from feet to fingertips. Enabling robots to perform full-range human-like movements from bending to reaching. Understands and reasons within the real, physical world. Gemini Robotics 2 pairs deep spatial reasoning with long-horizon planning, enabling robots to map multi-step sequences and complete complex, unfamiliar tasks. Supports multi-robot collaboration, allowing two robots to collaborate, and divide labor to complete a single task. Understands and responds to everyday commands. Gemini Robotics 2 can explain its approach while performing an action, while users can redirect it without using technical language. This makes it ideal for instructing robots through volatile, hazardous environments. See how Gemini Robotics handles a range of different tasks. Controls entire humanoid robots from feet to fingertips, translating intent into whole-body control to reach, bend, and balance. Controls complex humanoid hands and parallel grippers to unlock a new level of physical dexterity. Enables different types of robots to communicate and work together to solve complex workflows a single robot could not do alone. If you're interested in testing our models, please share a few details to join the waitlist. To ensure Gemini Robotics benefits humanity, we’ve taken a comprehensive approach to safety, from practical safeguards to collaborations with experts, policymakers, and our Responsibility and Safety Council. Supporting the new generation of physical AI We’re helping transform their discoveries into market-leading ventures – and to power an era of intelligent physical AI.