Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Harmfulness Evaluation on HarmBench

21.26Harmful Response Ratio

STAR-1

19.419231.844644.2756.6954Jan 30, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.01
21.26
2026.01
27.08
2026.01
32.39
2026.01
32.72
2026.01
40.2
2026.01
44.52
2026.01
45.35
2026.01
47.51
2026.01
51
2026.01
52.33
2026.01
52.33
2026.01
52.99
2026.01
52.99
2026.01
55.15
2026.01
56.98
2026.01
58.8
2026.01
59.3
2026.01
62.79
2026.01
65.28
2026.01
66.11
2026.01
67.28