Agent Security on A3S-BENCH Basic 1.0
45.27RTR@1 (%)Qwen3.5-35B
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Qwen3.5-35B2026.05 | 45.27 | 30.99 | 20.89 | 3.48 | |
| DeepSeek-V3.22026.05 | 37.54 | 23.78 | 14.15 | 3.78 | |
| Qwen3.5-122B2026.05 | 36.58 | 24.91 | 15.28 | 3.71 | |
| GLM-52026.05 | 35.85 | 23.12 | 13.37 | 3.81 | |
| MiniMax-M2.52026.05 | 32.39 | 21.33 | 12.9 | 3.95 | |
| Avg.2026.05 | 28.28 | 18.22 | 11.08 | 4.02 | |
| Qwen3.5-397B2026.05 | 27.48 | 17.76 | 10.76 | 4.08 | |
| Kimi-K2.52026.05 | 25.99 | 15.19 | 8.06 | 4.17 | |
| DeepSeek-V4-F2026.05 | 24.92 | 15.57 | 9.85 | 4.12 | |
| GPT-5.22026.05 | 8.83 | 4.41 | 3.12 | 4.55 | |
| Sonnet 4.52026.05 | 7.92 | 5.19 | 2.45 | 4.56 |