CWE-bench v2: Python
Cybersecurity · 2026-11-02
Pass@1 across the 35 Python tasks of CWE-bench v2's public and held-out sets, averaging four rollouts per task and rounded to a whole point by the publisher.
Benchmark source: Cwe Bench
Top models (higher is better)
| Model | Score | Source |
|---|---|---|
| GPT-6 Astra | 59.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Muse Wonder | 59.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Gemini 4 Argon | 57.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Claude Opus 5.5 | 54.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| DeepSeek V4.1 Flash 1017 | 50.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Grok 4.7 | 46.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Kimi K3.1 Pro | 46.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Qwen3.8 Max (0902) | 45.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Mistral Large 4 Preview | 44.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| GLM-5.3 | 39.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Hy4 Preview | 35.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |