Calibration

V* → DynaMath

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

DynaMath is estimated from V* with a offset logistic curve fitted on 5 models measured on both: y = 0.2360 + (0.8595 − 0.2360) / (1 + exp(−180.07·(x − 0.6578))), R² = 0.95, cross-validated error 3.3 pp. It is used for 1 estimate.

Estimated model	V*	DynaMath	Source
Step 3.7 Flash	95.3%	86.0%	estimated ± 3.3 pp, medium confidence