Google DeepMind has launched Gemini Robotics-ER 1.6, a new high-level reasoning model that significantly enhances robotic capabilities in spatial reasoning and multi-view understanding. Available through the Gemini API and Google AI Studio, this model introduces three core upgrades: improved pointing accuracy, multi-view success detection, and advanced instrument reading capabilities. These enhancements enable robots to perform precise object detection, integrate multiple camera feeds for task completion, and interpret various industrial instruments using agentic vision. Boston Dynamics has integrated Gemini Robotics-ER 1.6 into its Orbit AIVI-Learning product, allowing the Spot robot to autonomously patrol industrial sites and read data from instruments like pressure gauges. This integration, available since April 8, enhances Spot's performance in tasks such as visual inspection and liquid pooling detection. DeepMind highlights ER 1.6 as its safest model, with superior compliance in safety instruction and risk identification tests compared to previous models.