12:42 PM EDT, 03/12/2025 (MT Newswires) -- Alphabet's (GOOG, GOOGL) Google DeepMind unit said Wednesday it has introduced two new artificial intelligence models based on Gemini 2.0 designed specifically for robotics.

Gemini Robotics and Gemini Robotics-ER help robots understand and interact with the physical world more effectively, the company said.

Gemini Robotics is a vision-language-action model designed to control robots as it can adapt to different tasks and environments, allowing robots to handle objects and follow instructions in real time, it added.

Gemini Robotics-ER focuses on advanced spatial reasoning, enabling roboticists to integrate it into their own systems for direct control as it can understand object positions and generate safe movement paths, according to Google Deepmind.

Both models offer robust capabilities in generality, interactivity, and dexterity and can work across various robot types, from stationary robotic arms to humanoid robots in development, Google ( GOOG ) added.

The models integrate with existing low-level safety measures like collision avoidance and rely on Gemini's built-in features for higher-level understanding, the Google ( GOOG ) division said.

Google's ( GOOG ) unit added that it is partnering with robotics companies such as Apptronik, Agile Robots, and Boston Dynamics to support research and improvement in these models under everyday settings.

Price: 168.21, Change: +2.23, Percent Change: +1.34