Gemini Robotics 2 brings whole body intelligence to robots
DeepMind Blog · 2026-07-28
Google DeepMind releases Gemini Robotics 2, a suite of three AI models enabling humanoid robots to perform whole-body coordinated movement, advanced dexterous manipulation with five-fingered hands, and multi-robot collaboration, with on-device adaptation to new robot bodies in under two hours.
Appears in
Extraction
Topics: roboticsembodied-aihumanoid-robotsmulti-robot-collaborationphysical-ai
Claims
- Gemini Robotics 2 enables full-body humanoid control from feet to fingertips, allowing robots to walk, crouch, stretch, and manipulate objects in unstructured real-world environments.
- Gemini Robotics On-Device 2 can adapt to entirely new robot embodiments in just a few hours using fewer than 200 examples, running locally without network connectivity.
- The system enables multi-robot collaboration, allowing heterogeneous robots such as humanoids and bi-arm platforms to communicate and divide complex tasks that a single robot could not complete alone.
- A new ASIMOV-Agentic benchmark measures embodied reasoning agents' ability to refuse unsafe tool calls from lower-level action models and proactively request human intervention under uncertainty.
- Gemini Robotics ER 2 is publicly available via the Gemini API and Google AI Studio, while the VLA and On-Device models are limited to early-access partners.
Key quotes
Gemini Robotics 2 marks an important milestone on the path toward solving AGI in the physical world.
Gemini Robotics On-Device 2 is built specifically to handle these constraints — it is our most-efficient vision-language-action model (VLA) optimized to run locally on robotic devices.
By building this core intelligence, our goal is to enable AI in the physical world that can work alongside humans to solve complex challenges.