Calibration

AA ITBench → BrowseComp

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

BrowseComp is estimated from AA ITBench with a Hill curve fitted on 7 models measured on both: y = 0.0000 + (0.9427 − 0.0000)·x^6.00 / (0.31954^6.00 + x^6.00), R² = 0.32, cross-validated error 4.7 pp. It is used for 2 estimates.

Estimated model	AA ITBench	BrowseComp	Source
GLM-5.3-Flash	51.2%	89.0%	estimated ± 4.7 pp, low confidence
Grok 4.7	42.1%	79.1%	estimated ± 4.7 pp, low confidence