List your company

Illicit Task Completion on AgentHarm English prompts

72.7AgentHarm Score (AHS)STING
-0.6218.41537.4556.485Feb 18, 2026

Evaluation results

MethodLinks
STINGTarget Agent Model=Qwen3-Next, Smax=102026.0272.7
STINGTarget Agent Model=Qwen3-Next, Smax=52026.0267.8
STINGTarget Agent Model=DeepSeek-V3.2, Smax=102026.0261.8
STINGTarget Agent Model=DeepSeek-V3.2, Smax=52026.0257.2
STINGTarget Agent Model=Gemini 3 Flash, Smax=102026.0250.9
STINGTarget Agent Model=Gemini 3 Flash, Smax=52026.0247.6
Single-turn promptingTarget Agent Model=Gemini 3 Flash2026.0245.9
Single-turn promptingTarget Agent Model=Qwen3-Next2026.0235.1
STINGTarget Agent Model=GPT-5.1, Smax=102026.0234.1
STINGTarget Agent Model=Claude Sonnet 4.5, Smax=102026.0232.3
Single-turn promptingTarget Agent Model=DeepSeek-V3.22026.0231.2
STINGTarget Agent Model=GPT-5.1, Smax=52026.0229.7
STINGTarget Agent Model=Claude Sonnet 4.5, Smax=52026.0228
X-TeamingTarget Agent Model=Qwen3-Next, Smax=52026.0227
Single-turn promptingTarget Agent Model=GPT-5.12026.0224.3
Single-turn promptingTarget Agent Model=Claude Sonnet 4.52026.0216
X-TeamingTarget Agent Model=DeepSeek-V3.2, Smax=52026.0215.1
X-TeamingTarget Agent Model=Gemini 3 Flash, Smax=52026.0213.8
X-TeamingTarget Agent Model=GPT-5.1, Smax=52026.025
X-TeamingTarget Agent Model=Claude Sonnet 4.5, Smax=52026.022.2