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

Agent Safety Judgment research benchmarks · SOTA2 Research

Agent Safety Judgment

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

ROME Shortcut Decision-Making 1.0

94.53F1-Score

24

May 6, 2026

ROME Contextual Ambiguity 1.0

91.92F1 Score

24

May 6, 2026

ROME Implicit Risks 1.0 (IR)

86.81F1-Score

24

May 6, 2026

ROME Original 1.0 (Seed)

94.53F1-Score

24

May 6, 2026

Claude Sonnet 4.6

95.74F1 Score

4

May 6, 2026

ROME CA unsafe subset

Claude Sonnet 4.6

97.35F1 Score

4

May 6, 2026

ROME SDM unsafe

Claude Sonnet 4.6

100F1 Score

4

May 6, 2026

Popular datasets

Follow for updates

Join SOTA2 Contact