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

Safety Moderation research benchmarks – Page 2 · SOTA2 Research

Safety Moderation

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

Aegis Prompt 1.0

Roblox Guard 1.0

91.9F1 Score

7

Feb 26, 2026

95.7F1-Score

6

Jun 2, 2026

Beaver Response

84.4F1 Score

5

Feb 26, 2026

Nemo-Safety Response

0.835F1 Score

5

Feb 26, 2026

GuarEval Response

79.4F1 Score (Safety Moderation)

5

Feb 26, 2026

77.2F1 Score

5

Feb 26, 2026

Nemo-Safety Prompt

82F1 Score

5

Feb 26, 2026

GuardReasoner Safety Moderation Landscape (ToxicChat, OpenAI Mod, Aegis, SST, HarmBench, WildGuard, BeaverTails, StrongREJECT-Large, X-Sherpa, X-Refusal) 1:1 per-benchmark (test)

GuardReasoner-3B

78.2ToxicChat Score

4

Jun 26, 2026

Page 2 of 2

Popular datasets

Aegis XSTest Wild Guard GuarEval Nemo-Safety

Follow for updates

Join SOTA2 Contact