List your company

Multi-channel Attack Success Rate Evaluation on 98 attack cases (headline)

48Tool ASRGPT-OSS 120B
32.08836.21940.3544.481May 30, 2026

Evaluation results

MethodLinks
GPT-OSS 120BGroup=Agent-native, Parameters=120B2026.054824.523.5
Nemotron 3 Super 120BGroup=Agent-native, Parameters=120B2026.0544.921.423.5
Qwen3 Next 80BGroup=General, Parameters=80B, Behavioral Clustering=Agent-native2026.0543.919.424.5
Llama 3.3 70BGroup=General, Parameters=70B2026.0541.864.322.4
Kimi K2.5Group=Agent-native2026.0536.79.427.3
GLM 4.5 AirGroup=General, Series=Air2026.0532.751.518.9