CWE-bench v2 Pass@4
Cybersecurity · 2026-11-02
Share of the 250 CWE-bench v2 tasks solved in at least one of four one-hour high-reasoning rollouts according to the deterministic exploit and behaviour verifier.
Benchmark source: Cwe Bench
Top models (higher is better)
| Model | Score | Source |
|---|---|---|
| Claude Opus 5.5 | 79.2 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Muse Wonder | 78.8 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Gemini 4 Argon | 77.2 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| DeepSeek V4.1 Flash 1017 | 73.6 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| GPT-6 Astra | 72.8 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Grok 4.7 | 71.2 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Qwen3.8 Max (0902) | 64.8 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Hy4 Preview | 64.4 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Kimi K3.1 Pro | 64.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| GLM-5.3 | 63.6 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Mistral Large 4 Preview | 60.4 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |