List your company

Error Detection Calibration on Clean Benchmark 484 cases

48.3Empty Report RateSingle-Agent Detector
46.8256.8166.876.79May 12, 2026

Evaluation results

MethodLinks
Single-Agent DetectorArchitecture=Single-agent monolithic2026.0548.355.6
OptArgusArchitecture=Taxonomy-grounded multi-agent2026.0585.315.9