Harness-Bench

Harness-Bench is a benchmark study demonstrating that AI agent capability must be reported at the model-harness configuration level rather than attributed to the base model alone, providing a framework for more accurate, context-rich evaluation of co

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.