CWE-bench v1 - OWASP - A06 Insecure Design - Programmatic Pass@1: leaderboard
Collinear CWE-bench archived v1, OWASP - A06 Insecure Design - Programmatic Pass@1, reported as a percentage. 120 private audit-and-patch tasks; four rollouts per task at high reasoning, one hour each; timeouts graded on code at timeout. Native model-specific harnesses. Programmatic success requires exploit blocked and existing tests passing. Separate majority panel of three rubric-based agentic judges checks correctness, happy-path, generalization, integrity and soundness; any failed axis rejects a fix. Pass@1 and Pass@4 remain separate. This is archived Collinear v1, not the Artificial Analysis Stirrup two-hour implementation or CWE-bench v2. Source: https://cwe-bench.com/v1
Metric: Programmatic Pass@1 (%). Source: cwe-bench.com. Saturation forecast: Around August 2027. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Carbon | 69 | #2 |
| 2 | Gemini 4 Argon | 67 | #11 |
| 3 | Grok 4.6 | 54 | #44 |
| 4 | GPT-6 Astra | 52 | #6 |
| 5 | Claude Opus 5.5 | 52 | #5 |
| 6 | Grok 4.7 | 50 | #43 |
| 7 | Claude Fable 5.1 | 50 | #14 |
| 8 | Hy4 preview | 48 | #66 |
| 9 | Muse Spark 1.3 | 44 | #42 |
| 10 | DeepSeek V4.1 Flash | 42 | #58 |
| 11 | GPT-6 Sol | 35 | #24 |
| 12 | Inkling | 31 | #270 |
No result here: #1 Claude Fable 5.2, #10 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=cwe-bench-v1-owasp-a06-insecure-design-programmatic-pass-1 · How It Works · Data refreshed daily, snapshot 2026-11-12.