SpatialClaw: Training-Free Code-as-Action Interface Boosts VLM Agentic Spatial Reasoning

CMHungSteven · x · 2026-09-29

A new arXiv paper from an NVIDIA internship team studies how the action interface shapes VLM spatial reasoning agents:

Related event: NVIDIA Interns Release SpatialClaw: Code as Action Interface for VLM Spatial Reasoning(2 posts)→

Original post →

More from Research

Research channel →