Categories Companies Research

Loading the SOTA2 catalog…

Explore

Categories Companies Research

Contribute

Join SOTA2 Contact

SOTA2

© 2026 SOTA2

Build in Bellevue, USA

WizWand is now SOTA2 Research

Papers Benchmarks Datasets Tasks

Response Harmfulness Detection research benchmarks · SOTA2 Research

Response Harmfulness Detection

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

98.94F1 Score

100

Jun 1, 2026

GuardReasoner-Omni 4B

95.48Response Harmfulness F1

76

Jun 1, 2026

89.9F1 Score

59

Jun 1, 2026

72.1F1 Score

41

Jun 1, 2026

Response Harmfulness Detection Benchmarks (HarmBench, SafeRLHF, BeaverTails, XSTest, WildGuard)

0.833Macro Avg F1

21

May 29, 2026

GuardReasoner-Omni 4B

95.25F1 Score

5

Feb 26, 2026

GuardReasoner-Omni 2B

74.73F1 Score

5

Feb 26, 2026

Popular datasets

SafeRLHF XSTest HarmBench BeaverTails SPA-VL-Eval

Follow for updates

Join SOTA2 Contact