GigaBrain-0.7 Scales Embodied Foundation Models via Three-System Architecture
August 15, 2026
GigaBrain-0.7 utilizes a three-system architecture to unify understanding, prediction, and action in vision-language-action models. The model was pretrained on 37,000 hours of heterogeneous embodied data to improve generalization across diverse robot hardware.
HOW THIS AFFECTS YOU
●
researcherThis architecture offers a new template for scaling VLA models using heterogeneous robotic datasets.