Safety Evaluation on Advbench (test)
100PerformanceKimi-Audio-7B-Instruct
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Kimi-Audio-7B-InstructType=Audio, Size=7B2026.02 | 100 | — | — | — | |
| Eureka-Audio-InstructType=Ours, Size=1.7B2026.02 | 99.81 | — | — | — | |
| Qwen3-Omni-InstructType=Omni, Size=30B-A3B2026.02 | 99.61 | — | — | — | |
| Ming-Lite-Omni-1.5Type=Omni, Size=19B-A2.8B2026.02 | 99.23 | — | — | — | |
| Qwen2-AudioType=Audio, Size=7B2026.02 | 98.84 | — | — | — | |
| Audio Flamingo 3Type=Audio, Size=8B2026.02 | 98.26 | — | — | — | |
| Qwen2.5-Omni-7BType=Omni, Size=7B2026.02 | 96.92 | — | — | — | |
| MiniCPM-oType=Omni, Size=9B2026.02 | 95.76 | — | — | — | |
| Step-Audio-2-miniType=Audio, Size=8B2026.02 | 93.08 | — | — | — | |
| Qwen2.5-Omni-3BType=Omni, Size=3B2026.02 | 89.8 | — | — | — | |
| TRIDENT-EDGEModel Backbone=GEMMA-7B, Alignment Dataset=TRIDENT-EDGE2025.05 | — | 55 | 2.49 | 31 | |
| UnalignedModel Backbone=GEMMA-7B, Alignment Status=Unaligned2025.05 | — | 84 | 4.08 | 75 | |
| WILDBREAKModel Backbone=GEMMA-7B, Alignment Dataset=WILDBREAK2025.05 | — | 81 | 3.93 | 71 |