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

Reward Modeling Evaluation research benchmarks · SOTA2 Research

Reward Modeling Evaluation

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

Flexible Principles ScalarRM

85.3Chat Score

99

Jun 9, 2026

Reward Bench Factuality 2

Distribution-Calibrated Aggregation

56.6Pairwise Accuracy

64

Feb 26, 2026

RewardBench2 (test)

82.9Accuracy

20

Apr 6, 2026

Reward Bench Ties 2

Distribution-Calibrated Aggregation

91.8Pairwise Accuracy

12

Feb 26, 2026

Reward Bench Safety 2

Distribution-Calibrated Aggregation

72.3Pairwise Accuracy

12

Feb 26, 2026

Reward Bench Math 2

Distribution-Calibrated Aggregation

72.3Pairwise Accuracy

12

Feb 26, 2026

84.79Agreement

12

Feb 26, 2026

HelpSteer3 (test)

-5.89Score

5

May 12, 2026

UltraFeedback (test)

-3.12Score

5

May 12, 2026

SFT + TTL + DPO + TPO

88.1R-Bench Score

3

May 11, 2026

Popular datasets

Reward Bench RewardBench UltraFeedback RM-Bench HelpSteer

Follow for updates

Join SOTA2 Contact