Google Adds Gemini 4 Neon With Gemini 4 Hydrogen and 58 Score

Google adds Gemini 4 Neon with Gemini 4 Hydrogen, a faster robotics model that scores 58 and ships with a Robotics plugin in Google Antigravity.

Published
2 Min Read
Google Adds Gemini 4 Neon With Gemini 4 Hydrogen and 58 Score

Google introduced Gemini 4 Hydrogen last week, and it now pairs with a new Robotics plugin on Google Antigravity. Gemini 4 Neon is the launch label here, but the practical change is a faster robotics model with a benchmark score of 58.

- Advertisement -

For developers building robot applications, that means one workflow can now cover setup, calibration, simulation, and control. It also means Google is pushing the model into the part of robotics work where speed and integration matter more than a polished demo.

Gemini 4 Hydrogen and 58

Gemini 4 Hydrogen brings strong spatial and embodied reasoning to a much faster model, and it preserves the ability to drive a robot arm directly from a camera view and a plain-language goal. On the Artificial Analysis Robotics 0.1 benchmark, it controls a real 6-axis arm through eight everyday tabletop tasks, including pouring a drink and packing blocks into a box.

The score is 58, which places it similarly to Gemini 4 Argon while using less cost per token. That is the part developers will weigh in practice, because a model that is cheaper to run but only roughly matches another model changes budget decisions more than marketing language does.

Robotics plugin in Google Antigravity

The Robotics plugin packages work into skills Gemini can use on the user's behalf, and Google says it is built for hardware onboarding, simulation, and robot control. In one described setup, a developer can plug an I2RT YAM arm and an external depth camera into a laptop and ask Hydrogen in Antigravity to bring it up.

- Advertisement -

The plugin's skills install the software, find the camera, calibrate the arm, and run test motions, and the arm is ready for tasks in about 10 minutes. A similar flow also brings up a low-cost SO-ARM101 arm and an industrial grade FR3 Duo from scratch to live waving.

ERQA and developer choice

Google also says Gemini 4 Hydrogen brings the specialized skills of Gemini Robotics ER 2 into the core Gemini 4 model series, and it outperforms Gemini Robotics ER 2 on ERQA spatial reasoning tasks such as multi-view correspondence and 3D trajectory reasoning. That gives developers a clearer tradeoff: the newer model is pitched as faster and more efficient, yet the benchmark picture still shows a close race with Gemini 4 Argon rather than a clean leap ahead.

The open question is which hardware, software, or workload limits should push a developer toward Gemini 4 Hydrogen over Gemini 4 Argon or Gemini 4 Carbon. Google has given enough to start building, calibrating, and simulating inside Google Antigravity, but not enough to draw a sharp line between the three Gemini 4 models for every robotics job.

Advertisement
Share This Article
Technology analyst writing on semiconductors, cybersecurity, and Big Tech regulation. Holds a master's degree in Computer Science from MIT.