Calibration

AA EnterpriseOps-Gym → AA Harvey LAB

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

AA Harvey LAB is estimated from AA EnterpriseOps-Gym with a Hill curve fitted on 5 models measured on both: y = 0.0000 + (0.9470 − 0.0000)·x^6.00 / (0.21115^6.00 + x^6.00), R² = 0.98, cross-validated error 1.3 pp. It is used for 1 estimate.

Estimated model	AA EnterpriseOps-Gym	AA Harvey LAB	Source
GLM-5.3-Flash	33.2%	88.8%	estimated ± 1.3 pp, medium confidence