Google Releases Gemini 4 Argon, Advancing Robotics with Enhanced Physical Understanding and Control

Google has announced the release of Gemini 4 Argon for Robotics, a new AI model designed to significantly improve physical understanding, code generation, and visual-semantic reasoning in robotics. The model achieved an 87% score on the Embodied Reasoning (ERQA) benchmark, demonstrating its capability to solve physical and spatial problems in 3D environments. Argon is expected to accelerate progress toward physical AGI by enabling new workflows across the robotics stack, including robot control, real-to-simulation reconstruction, and iterative self-improvement on hardware.