Today's systems can handle limited spatial [music] tasks such as basic relationships or depth estimation. But they cannot robustly reason about spatial manipulation, 2D and 3D features, the relationships or operations [music] like mental rotation. This limits AI's ability to understand or interact with physical world.There is an opportunity to build large scale spatial reasoning models that treat geometry [music] and physical structure as first class primitives, not approximations layered on top of language ...