Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Human Preference Evaluation on EQ-Bench Creative and LongBench Write (60 pairs)

67.5Winrate

POLARIS

50.54854.94959.3563.751Jun 2, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.06
67.557.52022.5
2026.06
51.23532.532.5