CWE-bench v1: Insecure Design
Code · 2026-09-25
Programmatic pass@1 on the OWASP 2025 A06 subset of CWE-bench v1; exact category denominator is not published.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 4 Argon | 67.0 |
| Grok 4.6 | 54.0 |
| Claude Opus 5.5 | 52.0 |
| GPT-6 Astra | 52.0 |
| Claude Fable 5.1 | 50.0 |
| Grok 4.7 | 50.0 |
| Hy4 Preview | 48.0 |
| Muse Spark 1.3 | 44.0 |
| DeepSeek V4.1 Flash | 42.0 |
| GPT-6 Sol | 35.0 |
| Inkling | 31.0 |