Malicious code generation audit on Innoc2Scam-bench C2
1,035Promptsgemini-2.5-pro
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| gemini-2.5-pro2025.09 | 1,035 | 631 | 380 | 143 | |
| gpt-52025.09 | 1,035 | 933 | 18 | 213 | |
| claude-sonnet-42025.09 | 1,035 | 940 | 83 | 370 | |
| grok-code-fast-12025.09 | 1,035 | 1,018 | 13 | 452 | |
| gemini-2.5-flash2025.09 | 1,035 | 1,017 | 1 | 468 | |
| qwen3-coder2025.09 | 1,035 | 1,029 | 1 | 486 | |
| deepseek-chat-v3.12025.09 | 1,035 | 1,024 | 6 | 505 |