DeepMind's Gemini Robotics 2 moves humanoid control from the arms to the whole body

Google DeepMind announced Gemini Robotics 2 on July 30, 2026 — three models spanning a vision-language-action model that converts vision and language into motor control for whole-body humanoid movement (walking, crouching, manipulating while reasoning through tasks), an embodied-reasoning agent model for multi-step plans lasting minutes, and an on-device variant DeepMind says adapts to a new robot body with a few hours of data. The prior generation controlled mainly the upper body; this one directs the full humanoid and supports multi-robot collaboration.

What it means Whole-body control plus few-hour adaptation to new hardware is what turns humanoid demos into deployable platforms — the robotics stack is consolidating around foundation models.