List your company

Writing and Arena Evaluation on ArenaHard v2

89ArenaHard-v2 Creative AccuracyRDPO
69.86474.83279.884.768May 13, 2026

Evaluation results

MethodLinks
RDPO2026.058976.1
GRPO2026.0582.565.8
RL initialization model2026.0570.649.4