CWE-bench v1: Rust
Code · 2026-09-25
Programmatic pass@1 across the three Rust tasks in CWE-bench v1, averaging four rollouts per task.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 4 Argon | 92.0 |
| Claude Fable 5.1 | 67.0 |
| Claude Opus 5.5 | 67.0 |
| Grok 4.7 | 67.0 |
| Grok 4.6 | 50.0 |
| DeepSeek V4.1 Flash | 42.0 |
| GPT-6 Astra | 42.0 |
| Hy4 Preview | 42.0 |
| Inkling | 42.0 |
| Muse Spark 1.3 | 42.0 |
| GPT-6 Sol | 25.0 |