Google DeepMind's Gemini Robotics 2 Series redefines humanoid robot control

Google DeepMind has unveiled the **Gemini Robotics 2 series**, a next-generation family of models designed to supercharge humanoid robot control. Announced on July 30, 2026, the series introduces two flagship models: **Gemini Robotics ER 2** for embodied reasoning and **Gemini Robotics On-Device 2** for efficient onboard inference. These models are engineered to tackle complex, real-world tasks by enabling multi-robot collaboration and intelligent whole-body control, addressing one of the biggest challenges in robotics today.

The ER 2 model focuses on high-level reasoning, allowing robots to plan and execute multi-step tasks with minimal human intervention. Meanwhile, the On-Device 2 model prioritizes efficiency, running inference directly on the robot’s hardware to reduce latency and dependency on cloud connections. Together, these models promise to bridge the gap between theoretical AI and practical, deployable robotics, unlocking new possibilities for automation in dynamic environments.