Gemini Robotics

Gemini Robotics

Bringing AI into the Physical World

168 followers

Visit website
Gemini Robotics from Google Deepmind, is the Gemini 2.0-based AI models for robots. Multimodal, general, interactive, and dexterous. Powers ALOHA 2, Apptronik Apollo, and more.
This is the 3rd launch from Gemini Robotics. View more
Gemini Robotics 1.5

Gemini Robotics 1.5

Bringing AI agents to the physical world
Gemini Robotics 1.5 is a new agentic framework from Google that enables robots to perceive, plan, and act. It combines a reasoning model for complex planning and tool use with a vision-language-action model for physical execution.
Gemini Robotics 1.5 gallery image
Gemini Robotics 1.5 gallery image
Launch tags:Robots•Artificial Intelligence
Launch Team
NMI Payments
NMI Payments
Don’t Integrate Payments Until You Read This Guide
Promoted

What do you think? …

Zac Zuo
Flowtica Scribe
Hunter
📌
Hi everyone! The biggest challenge in robotics has always been bridging the gap from simple, pre-programmed tasks to handling the complexity of the real world. A robot that can put a block in a box is one thing, a robot that can figure out which box based on local recycling rules is another. The new Gemini Robotics 1.5 is so interesting. It's not a single model, but an agentic framework. It pairs a reasoning model (ER 1.5) that acts as the "brain": planning, reasoning, and even using Google Search—with an action model (VLA 1.5) that executes the physical tasks. The "thinking before acting" capability is a big deal, but the most impressive part for me is its ability to learn across different embodiments. Seeing a skill transfer from a simple robot arm to a full humanoid without retraining is a major step towards truly general-purpose robotics.