Benchmark
CyberSecEval 2
Meta's cybersecurity evaluation suite measuring models' propensity to assist with cyberattacks, generate insecure code, and support offensive security operations. Lower scores indicate lower cyber risk.
What it measures
Cybersecurity risk across insecure coding assistance, cyberattack facilitation, prompt injection susceptibility, and offensive security capability categories
Safety areas: Cyber safety
Score direction: lower_is_better
Public model scores
No comparable public model scores found yet.
Primary source
CyberSecEval 2: A Wide-Ranging Cybersecurity Evaluation Suite for Large Language ModelsStatus: Verified. Last verified: 2026-06-13.