Artificial intelligence news.
Straight from the original announcements.

Google launches Gemini 4 Hydrogen for faster general-purpose robotics

November 17, 2026By the AI News Bites Editorial TeamSource: Google for Developers
Google launches Gemini 4 Hydrogen for faster general-purpose robotics
Image: Google for Developers

Google has launched Gemini 4 Hydrogen, a faster model for general-purpose robotics, alongside a Robotics plugin for its Google Antigravity development environment. The company says the combination can take developers from a robot in a box to a working application in an afternoon.

Key points
  • Gemini 4 Hydrogen is a faster model for general-purpose robotics.
  • Google reports a score of 58 on the Artificial Analysis Robotics 0.1 benchmark.
  • The Antigravity Robotics plugin adds hardware onboarding, MuJoCo simulation and robot control skills.
  • Google says a YAM arm can be ready for tasks in about 10 minutes.
  • Hydrogen is available through Antigravity, the Gemini API and Google AI Studio.

A faster model for robot control

Gemini 4 Hydrogen is designed to combine general-purpose language abilities with spatial and embodied reasoning. Google says the model can accept a goal in plain language and a camera view, then plan a task, write and run control code, and adjust its actions until the job is complete.

On the Artificial Analysis Robotics 0.1 benchmark, Google says Hydrogen controlled a real 6-axis arm through eight everyday tabletop tasks, including pouring a drink and packing blocks into a box. It scored 58, which the company describes as similar to Gemini 4 Argon at a lower cost per token.

Google also says Hydrogen brings capabilities from its Gemini Robotics ER 2, or Embodied Reasoning, models into the core Gemini 4 series. On the public ERQA benchmark, the company says Hydrogen outperformed Gemini Robotics ER 2 on spatial reasoning tasks including multi-view correspondence and 3D trajectory reasoning.

Antigravity adds robotics tools

The Robotics plugin for Google Antigravity is intended to combine the software needed to build and test robot applications. Its skills can install drivers, discover cameras, calibrate an arm and verify safe motion. They can also create and preview MuJoCo simulation scenes, identify objects, plan grasps and control a robot in a closed loop.

Google says developers can use the environment to generate controllers and simulations, then develop applications through a Real-to-Sim-to-Real workflow. Simulation assets can be described in prompts, imported from links or documents, or created from workspace photos. Developers can also adjust shapes, positions and physical properties through chat to test reaching and grasping before using physical hardware.

The plugin supports several example workflows. Google says an I2RT YAM arm can be calibrated and made ready for tasks in about 10 minutes from a USB connection. It also describes a snack-sorting conveyor application that went from a whiteboard idea to a working sorter in about an hour, as well as a soap storefront using two YAM Pro arms for order packing.

How developers can get started

In Google Antigravity, users can select Gemini 4 Hydrogen, identified as gemini-4-hydrogen, then install the Robotics plugin by opening Customizations and searching for Robotics on the Build With Google shelf. Google suggests trying the prompt, “Bring up my SO-101 arm and make it wave.”

Google Antigravity is available to paid Google AI Pro or Ultra subscribers, according to the announcement. Developers can also use Gemini 4 Hydrogen through the Interactions API, with day-one support in the google-genai SDKs, or try spatial prompts, live camera streams and structured tool calls in Google AI Studio.

The launch matters because Google is targeting the practical work surrounding robotics, including hardware setup, calibration, simulation and deployment, rather than presenting the model only as a reasoning system. The company says this packaging is intended to make everyday robot application development faster and more accessible.

Related news

Google gives Gemini the ability to handle calls, bills and shopping
Products and toolsNovember 17, 2026, 18:00

Google gives Gemini the ability to handle calls, bills and shopping

Google is expanding Gemini beyond chat with task features that can schedule appointments, analyze bills, shop for products and manage grocery orders. The update is…

Gemini 4 Hydrogen ranks third in Ultralytics auto-annotation test
ResearchNovember 16, 2026, 17:22

Gemini 4 Hydrogen ranks third in Ultralytics auto-annotation test

Gemini 4 Hydrogen ranked third among 46 vision-language models in Ultralytics’ auto-annotation benchmark, scoring 0.553 mAP 50 across 36 object detection datasets. The…

Salesforce Builds Interactive Testbed for Real-Time Visual AI Assistants
ResearchNovember 12, 2026, 17:00

Salesforce Builds Interactive Testbed for Real-Time Visual AI Assistants

Salesforce AI Research has built Interactive Visual Gym, a simulated environment for testing whether real-time visual AI assistants can change the outcome of physical…