← All benchmarks

CTF-Archive Diamond (IRT-imputed estimate)

Agents · 2026-05-01

CAISI's Item Response Theory estimate of CTF-Archive Diamond success rate, imputed from a subset of evaluated samples. The target benchmark contains 285 challenges, but these estimates are not direct outcomes over that complete cohort and no numeric imputation uncertainty is published. Retained as diagnostic context only.

Top models (higher is better)