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

Multi-turn Chat Evaluation on MT-Bench

8.58MT-Bench Score

VRM-PPO

4.0045.1926.387.568Mar 5, 2026Mar 21, 2026Apr 6, 2026Apr 22, 2026May 8, 2026May 24, 2026Jun 9, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.03
8.58--
2026.03
8.44--
2026.06
8.018.67.43
2026.03
7.98--
2026.06
7.988.67.2
2026.03
7.97--
2026.03
7.86--
2026.06
7.868.57.23
2026.06
7.858.67.1
2026.06
7.858.487.21
2026.06
7.848.667.01
2026.03
7.83--
2026.06
7.828.617.02
2026.03
7.81--
2026.03
7.81--
2026.06
7.88.557.05
2026.06
7.778.57.04
2026.03
7.74--
2026.06
7.738.37.15
2026.03
7.64--
2026.03
7.63--
2026.06
7.598.236.96
2026.03
7.58--
2026.06
7.237.646.83
2026.06
7.217.686.74
2026.06
7.167.626.75
2026.06
7.157.356.95
2026.06
7.137.646.63
2026.06
7.117.66.62
2026.06
7.057.556.6
2026.06
7.057.516.58
2026.06
7.017.496.53
2026.06
6.997.466.52
2026.06
6.627.196.05
2026.03
6.55--
2026.03
6.38--
2026.03
6.24--
2026.03
6.13--
2026.04
5.28--
2026.04
4.88--
2026.04
4.4--
2026.04
4.18--