AsianFin — At the GTC 2025 global conference, ManyCore, one of Hangzhou's "Six Little Dragons," announced the open-sourcing of its self-developed spatial understanding model, SpatialLM. This model can generate physically accurate 3D scene layouts from just a single video.
SpatialLM breaks through the limitations of traditional large language models in understanding the geometry and spatial relationships of the physical world, enabling machines to possess spatial cognition and reasoning capabilities. It is expected to provide a foundational training framework for spatial understanding in fields such as embodied intelligence.
Explore more exclusive insights at nextfin.ai.
