Benchmarks / Vulnerability research

Vulnerability research

Find, reproduce, or exploit vulnerabilities, for example CTF challenges or CVEs.

Part of Security.

• CVE-Benchby UIUCTask level · Success rate (one-day, 5 attempts)

  12.5%AutoGPT with GPT-4o · Mar 2025

  Open
• Cybenchby Stanford UniversityTask level · Unguided % solved

  100%Claude Mythos Preview · 7 Apr 2026

  Saturated
• CyberGymby UC BerkeleyTask level · Success rate (Level 1)

  98.5%Creation (天工), multi-model · 7 Sep 2026

  Saturated