CWE-bench v2: Error Handling and Logic Flaws
Cybersecurity · 2026-11-02
Pass@1 on the 17 tasks of Collinear AI's K10 weakness class in CWE-bench v2, rounded to a whole point by the publisher. The ten classes are the publisher's own grouping, not an official taxonomy.
Benchmark source: Cwe Bench
Top models (higher is better)
| Model | Score | Source |
|---|---|---|
| Gemini 4 Argon | 71.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Muse Wonder | 66.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| GPT-6 Astra | 59.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Claude Opus 5.5 | 56.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Grok 4.7 | 53.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Kimi K3.1 Pro | 44.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| DeepSeek V4.1 Flash 1017 | 43.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Mistral Large 4 Preview | 40.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Hy4 Preview | 37.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| GLM-5.3 | 35.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |
| Qwen3.8 Max (0902) | 28.0 | CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI |