Calibration

CursorBench 3.1 → SWE-bench Pro

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

SWE-bench Pro is estimated from CursorBench 3.1 with a Michaelis–Menten + offset curve fitted on 5 models measured on both: y = 0.0000 + 2.0000·x / (1.19553 + x), R² = 0.71, cross-validated error 6.9 pp. It is used for 0 estimates.