NewsBite analysis · Robotics
The Decoder
Google Deepmind unveils Gemini Robotics 2 to power robots of all shapes from tabletop arms to humanoids
Google Deepmind unveiled Gemini Robotics 2, its most sophisticated vision-language-action (VLA) model to date, aimed at powering a wide range of robotic systems from tabletop arms to humanoid robots. The introduction of this model signifies Deepmind's commitment to enhancing robotic functional capabilities through advanced perception and decision-making. Developers can apply for early access via a waitlist to leverage these innovations.
MonitorConfidence: HighShort term horizon
Published 31 Jul 2026, 20:10 UTC1 min read
Visit original article ↗Lead image — The Decoder
Summary
TL;DR
- Gemini Robotics 2 is Deepmind's latest VLA model.
- It supports a variety of robots, from small arms to humanoids.
- The model combines image recognition, language processing, and action control.
- Developers can apply for early access through a waitlist.
- Gemini Robotics ER 2 is also introduced for higher-level control.
Interesting facts
- Gemini Robotics 2 acts as an 'intelligence layer' for robots.
- The model can coordinate multiple robots simultaneously.
- It enhances the physical operation and interaction capabilities in various environments.