Refusal Evaluation
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
99.6Toxic Refusal Rate
40
Apr 15, 2026
51ASR
40
Apr 15, 2026
100Refusal Rate
25
Apr 21, 2026
98XS
15
Mar 5, 2026
95.5Refusal Rate
15
Feb 26, 2026
92.35Reject Rate
13
Feb 26, 2026
95.21Refusal Rate
7
Mar 17, 2026
85.31Refusal Rate
7
Mar 17, 2026
94.66Refusal Rate
7
Mar 17, 2026
89.45Refusal Rate
7
Mar 17, 2026
84.35Refusal Rate
7
Mar 17, 2026
96.1Refusal Rate
7
Mar 17, 2026
99Not Unsafe Rate
2
May 1, 2026
71Baseline Performance
1
Feb 26, 2026
61.27Refusal Rate (Before)
1
Feb 26, 2026