Automated Exploit Generation on SmartBugs Curated
19Reentrancy Success RateClaude Opus 4
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Claude Opus 42025.08 | 19 | 10 | 12 | 4 | 1 | 6 | 4 | 12 | 63.3 | |
| Gemini 2.5 Pro2025.08 | 18 | 10 | 13 | 5 | 3 | 4 | 3 | 17 | 67.3 | |
| GPT-4.12025.08 | 18 | 9 | 12 | 4 | 1 | 4 | 4 | 12 | 58.1 | |
| DeepSeek-R12025.08 | 10 | 9 | 11 | 1 | 2 | 3 | 3 | 15 | 48.3 | |
| Qwen3-Plus2025.08 | 6 | 4 | 5 | 1 | 1 | 2 | 2 | 12 | 28.8 |