Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Safety Evaluation on BBQ
Loading...
9.21
Harmful Content Rate (HCRate)
DeepSeek-V3
-1.2112
69.1319
139.475
209.8181
Jun 4, 2026
Harmful Content Rate (HCRate)
Non-Response Rate (NRRate)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Harmful Content Rate (HCRate)
Non-Response Rate (NRRate)
DeepSeek-V3
Attack=None
2026.06
9.21
78
GPT-4o
Attack=None
2026.06
13.16
59.27
Claude-3.7-Sonnet
Attack=None
2026.06
13.16
43.82
GHOSTWRITER
Target Model=GPT-4o, A...
2026.06
76.32
81.64
GHOSTWRITER
Target Model=Claude-3....
2026.06
102.63
73.45
GHOSTWRITER
Target Model=Claude-3....
2026.06
103.95
87.82
GHOSTWRITER
Target Model=GPT-4o, A...
2026.06
139.47
83.45
GHOSTWRITER
Target Model=DeepSeek-...
2026.06
176.32
96
GHOSTWRITER
Target Model=DeepSeek-...
2026.06
269.74
97.64
Feedback
Search any
task
Search any
task