Google: Gemini Robotics-ER 1.6 Brings Instrument Reading and Spatial Understanding

On April 14, Google introduced Gemini Robotics-ER 1.6 through the Gemini API changelog — a new version of the model specialized for robotics and physical interaction with the world.

New Capabilities

The model brings three key improvements:

• Instrument reading — the ability to interpret physical measuring devices, scales, and displays
• Improved spatial understanding — better perception of three-dimensional space and relationships between objects
• Advanced physical reasoning — more accurate predictions of physical interactions and consequences of actions

Migration Deadline

The previous version, Gemini Robotics-ER 1.5, will be shut down on April 30, 2026 at 9:00 AM PST. Development teams using version 1.5 must migrate within two weeks.

The Bigger Picture

This release continues Google’s focus on “physical AI” — a field where generative models move beyond the digital world into physical reality. Combined with NVIDIA’s RoboLab benchmark (released last week), the industry is rapidly building infrastructure for autonomous robots that understand and manipulate physical environments.