Calibration

V* → MLVU (M-Avg)

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

MLVU (M-Avg) is estimated from V* with a Michaelis–Menten curve fitted on 5 models measured on both: y = 0.9641·x / (0.11439 + x), R² = 0.52, cross-validated error 2.2 pp. It is used for 1 estimate.

Estimated model	V*	MLVU (M-Avg)	Source
Step 3.7 Flash	95.3%	86.1%	estimated ± 2.2 pp, medium confidence