Long-query Safety on Long-query Safety Suite Chinese
99.2Operational Data PerformanceOyster-II
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Oyster-II2026.07 | 99.2 | 97.05 | 98.93 | 99.2 | 99.2 | 98.39 | 98.93 | 99.2 | 92 | 95.7 | 97.78 | |
| Qwen3.5-397B2026.07 | 97.86 | 40.75 | 94.1 | 99.2 | 47.72 | 84.99 | 79.36 | 98.39 | 94 | 95.9 | 83.23 | |
| Oyster-I2026.07 | 93.57 | 69.71 | 90.08 | 93.57 | 83.65 | 86.06 | 70.51 | 93.83 | 84.71 | 89.8 | 85.55 | |
| Qwen3-14B2026.07 | 82.04 | 14.48 | 58.45 | 84.72 | 21.18 | 57.1 | 33.78 | 54.16 | 70.94 | 78.6 | 55.54 | |
| Qwen3-Max2026.07 | 54.06 | 21.18 | 84.99 | 83.38 | 20.64 | 30.83 | 63.81 | 45.04 | 80.12 | 88.6 | 57.27 |