List your company

Reward Model Evaluation on Text-Coding (test fraction 0.2)

80.9Pair AccuracyArena BT
58.22864.1147075.886Jun 24, 2026

Evaluation results

MethodLinks
Arena BTReward Model=proprietary RM2026.0680.90.77310.851
Arena Avg RewardReward Model=proprietary RM2026.0680.50.7790.9940.851
DUALEVALReward Model=proprietary RM2026.0677.80.9590.8780.845
Static 2PL2026.0673.110.7730.763
Arena BTReward Model=Skywork-Reward-V2-Qwen3-8B2026.06680.2550.4160.511
Arena Avg RewardReward Model=Skywork-Reward-V2-Qwen3-8B2026.0667.30.1970.3110.432
DUALEVALReward Model=Skywork-Reward-V2-Qwen3-8B2026.0659.10.9760.8050.789