Exploit Generation on Path Traversal (PT) Vulnerabilities (input level 5)
18.53Total Scoreo3
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| o32026.06 | 18.53 | 81 | 11 | 3.94 | 1.17 | 2.52 | 3.22 | 3.59 | 3.16 | |
| Qwen3-235B2026.06 | 16.12 | 36 | 0 | 3.91 | 1 | 2.28 | 3.23 | 2.65 | 2.53 | |
| Claude Sonnet 4.52026.06 | 15.77 | 65 | 3 | 3.17 | 1.06 | 2.25 | 3.02 | 2.8 | 2.8 | |
| DeepSeek-V3.22026.06 | 15.18 | 69 | 5 | 3.47 | 1 | 2.13 | 2.64 | 2.66 | 2.33 | |
| Kimi-K2.52026.06 | 15.09 | 56 | 4 | 3.57 | 1 | 2.16 | 3.17 | 2.84 | 2.76 | |
| Qwen3-8BFine-tuning strategy=Base2026.06 | 13.53 | 42 | 0 | 3.49 | 0.72 | 1.98 | 2.68 | 2.16 | 2.06 | |
| Llama4-Maverick-400B2026.06 | 12.46 | 36 | 0 | 3.34 | 0.73 | 1.91 | 2 | 2.07 | 2.13 | |
| Qwen3-8BFine-tuning strategy=SFT2026.06 | 10.18 | 55 | 1 | 2.64 | 0.73 | 1.48 | 1.89 | 2.1 | 1.46 |