CWE-bench v2 Held-Out Set Pass@4
Cybersecurity · 2026-11-02
Share of the 150 held-out CWE-bench v2 tasks solved in at least one of four rollouts.
Benchmark source: Cwe Bench
Top models (higher is better)
| Model | Score | Source |
|---|---|---|
| Claude Opus 5.5 | 80.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Muse Wonder | 78.7 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| DeepSeek V4.1 Flash 1017 | 74.7 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Gemini 4 Argon | 74.7 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| GPT-6 Astra | 74.7 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Grok 4.7 | 69.3 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Qwen3.8 Max (0902) | 66.7 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Kimi K3.1 Pro | 64.7 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| GLM-5.3 | 62.7 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Hy4 Preview | 62.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Mistral Large 4 Preview | 61.3 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |