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

Harmlessness Evaluation on HarmBench

76.6Win Rate

fsd-var

22.62436.63750.6564.663Mar 11, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.03
76.610.612.7
2026.03
75.312.312.4
2026.03
74.712.612.7
2026.03
72.915.711.4
2026.03
7114.814.1
2026.03
68.120.411.5
2026.03
67.313.419.3
2026.03
67.17.825.1
2026.03
39.628.631.8
2026.03
3822.239.8
2026.03
34.23728.9
2026.03
30.448.920.7
2026.03
28.646.524.9
2026.03
2852.219.9
2026.03
24.749.725.5