Last released Aug 24, 2026
A process-quality benchmark for LLM agents: it grades how an agent worked, not whether it passed — and every LLM-as-a-judge sits a calibrated exam before it may score.
Supported by
