Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Agentic Safety Moderation on AgentHazard
Loading...
109
ASR
w/o guardrail
68.44
78.97
89.5
100.03
May 28, 2026
ASR
∆ASR
TTFT (mean)
TTFT (p95)
TPOT (mean)
TPOT (p95)
Completion Tokens (Avg)
Updated 5d ago
Evaluation Results
Method
Method
Links
ASR
∆ASR
TTFT (mean)
TTFT (p95)
TPOT (mean)
TPOT (p95)
Completion Tokens (Avg)
w/o guardrail
Guardrail=None
2026.05
109
-
-
-
-
-
-
Qwen3Guard-Gen-4B
Guardrail=Qwen3Guard-G...
2026.05
109
0
0.1208
0.3226
0.0127
0.0129
13.02
Llama-Guard-3-8B
Guardrail=Llama-Guard-...
2026.05
109
0
0.1008
0.3264
0.0118
0.0124
3
AgentDoG 1.5-0.8B
Base Model=Qwen3.5 Thi...
2026.05
76
-12.69
0.1892
0.4603
0.0148
0.0149
358.44
AgentDoG 1.5-4B
Base Model=Qwen3.5 Thi...
2026.05
70
-15
0.296
0.7378
0.0207
0.0209
493.38
Feedback
Search any
task
Search any
task