Back to Search
G

Gemini Robotics 1.5

Model

Gemini Robotics 1.5 is Google DeepMinds multitask VLA model that extends Gemini-LLMs capabilities into the physical world. It supports cross-embodiment zero-shot generalization, enabling a single model to control different robot platforms (ALOHA 2, Bi-arm Franka, Apptronik Apollo) without per-robot fine-tuning. The model performs visuomotor control, task planning, and closed-loop reasoning for real-world manipulation tasks.

Details

Updated:7/7/2026
open sourcefalse
release date2025-09-25
paper urlhttps://storage.googleapis.com/deepmind-media/gemini-robotics/Gemini-Robotics-1-5-Tech-Report.pdf
model familyGemini Robotics

Tags

VLACross-EmbodimentVision-Language-ActionGoogle DeepMindRobot Manipulation

Relationships

Sources

Google DeepMind - Gemini Robotics 1.5 Blog
website
Visit
Gemini Robotics 1.5 Technical Report
website
Visit

Appears In

No related landscapes.