Safety Defense
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
6.52ASR
7
May 28, 2026
7.88ASR
7
May 28, 2026
12.58ASR
7
May 28, 2026
10ASR
7
May 28, 2026
0.4Training Time (hr)
5
May 15, 2026
1Inference Overhead (Multiplier)
4
May 15, 2026
100Hate Rate
4
Mar 24, 2026