benchgap
Calibration

ExploitBench Internal Port → ExploitBench

ExploitBench is estimated from ExploitBench Internal Port with a Michaelis–Menten curve fitted on 7 models measured on both: y = 1.0591·x / (0.02302 + x), R² = 0.97, cross-validated error 7.0 pp. It is used for 0 estimates.

Fitted on

ModelExploitBench Internal PortExploitBench
GPT-5.6 Sol5.5%73.5%
GPT-6.1 Sol21.5%99.7%
GPT-6.1 Terra24.5%97.1%
GPT-6.2 Astra53.5%100.0%
GPT-6.2 Sol50.5%100.0%
GPT-6 Astra39.0%100.0%
GPT-6 Cyber42.0%100.0%