GoogleAlex Gruenstein, Gabe Cohen, Gemini Robotics TeamAnnouncement
Gemini 4 Hydrogen: a faster model for general purpose robotics
Summary
Google introduced Gemini 4 Hydrogen, a faster and more efficient model for general-purpose robotics that preserves the ability to directly control robot arms from camera images and plain-language goals. On the Artificial Analysis Robotics 0.1 benchmark, Hydrogen scores 58 on eight everyday tabletop tasks, similar to Argon at a lower cost per token. The model outperforms Gemini Robotics ER 2 on the public ERQA benchmark for spatial reasoning tasks such as multi-view correspondence and 3D trajectory reasoning. Google also launched the Robotics plugin for Google Antigravity, which packages robot bring-up, MuJoCo simulation, perception, and application building into skills Gemini can use. A developer can go from a robot in a box to a working application in about an afternoon, such as building a snack-sorting conveyor in an hour.
Context
Gemini 4 Argon and Carbon previously demonstrated frontier physical reasoning for robotics, but Gemini 4 Hydrogen is a faster, more efficient model that preserves the ability to directly control robot arms from camera images and plain-language goals.
Approach / What changed
The Robotics plugin for Google Antigravity packages robot bring-up, MuJoCo simulation, perception and control, and application building into skills that Gemini 4 Hydrogen can use on behalf of a developer, enabling end-to-end real-to-sim-to-real development from a single prompt.
Takeaways
Gemini 4 Hydrogen scores 58 on the Artificial Analysis Robotics 0.1 benchmark, controlling a real 6-axis arm through eight tabletop tasks at a lower cost per token than Argon.
The Antigravity Robotics plugin provides skills for robot bring-up, MuJoCo simulation, perception and control, and application building, enabling real-to-sim-to-real development from a single prompt.
A developer can set up a YAM arm from USB connection to calibrated and ready in about 10 minutes, and build a working application like a snack-sorting conveyor in about an hour.
FAQ
How does Gemini 4 Hydrogen perform on robotics benchmarks?
Gemini 4 Hydrogen scores 58 on the Artificial Analysis Robotics 0.1 benchmark, controlling a real 6-axis arm through eight everyday tabletop tasks such as pouring a drink and packing blocks into a box. This is similar to Argon's score but at a lower cost per token.
What does the Antigravity Robotics plugin include?
The Robotics plugin for Google Antigravity provides skills for robot bring-up (installing drivers, calibrating the arm), MuJoCo simulation inside Antigravity, perception and closed-loop control using Gemini's spatial understanding, and application building that generates the controller and simulation code.
How quickly can a developer set up a robot with this system?
A developer can go from a robot in a box to a working application in about an afternoon. For example, a snack-sorting conveyor application was built from a whiteboard idea to a working sorter in about an hour, and a YAM arm can be calibrated and ready for tasks in about 10 minutes.
Source and freshness
- Source
- developers.googleblog.com
- Published
- Added to Yomu
Summaries are generated from the original by a language model. Check the original before relying on it.
Keep reading
Related reading
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
The best engineering reads, once a week
One useful email. Unsubscribe anytime.