Calibration

HLE w/o tools → CritPt

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

CritPt is estimated from HLE w/o tools with a linear curve fitted on 31 models measured on both: y = 0.6049·x + -0.0802, R² = 0.68, cross-validated error 6.2 pp. It is used for 1 estimate.

Estimated model	HLE w/o tools	CritPt	Source
Pareto 26.9	49.0%	21.6%	estimated ± 6.2 pp, medium confidence