benchgap
Calibration

CursorBench 4.0 → FrontierCode 1.1 Main

FrontierCode 1.1 Main is estimated from CursorBench 4.0 with a Michaelis–Menten curve fitted on 5 models measured on both: y = 0.6925·x / (0.20100 + x), R² = 0.38, cross-validated error 5.7 pp. It is used for 0 estimates.

Fitted on

ModelCursorBench 4.0FrontierCode 1.1 Main
Claude Haiku 5.548.4%46.4%
Claude Opus 546.6%53.4%
Claude Opus 5.557.8%54.4%
Claude Sonnet 534.1%42.7%
Claude Sonnet 5.555.5%46.2%