OSReward

OSReward is a human-gold benchmark developed to evaluate VLM judges scoring computer-use agents, providing a reference standard for assessing agent performance. It serves as a critical tool for measuring and comparing the accuracy of vision-language

Two-hop subgraph: this entity, every entity it directly relates to, and every entity those neighbors relate to. Drag a node, scroll to zoom, click to inspect — or click any neighbor and re-center the atlas there.