Calibration

OSWorld 2.0 → GDPval-AA

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

GDPval-AA is estimated from OSWorld 2.0 with a Michaelis–Menten curve fitted on 20 models measured on both: y = 0.5957·x / (0.05211 + x), R² = 0.71, cross-validated error 7.5 pp. It is used for 1 estimate.

Estimated model	OSWorld 2.0	GDPval-AA	Source
Claude Opus 4.7	13.9%	43.3%	estimated ± 7.5 pp, medium confidence