Multi-turn Instruction Following
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
9.16MT-Bench Score (GPT-4)
137
Jun 11, 2026
92.31CSR
38
Apr 21, 2026
7.54Reward Score
26
Mar 10, 2026
89.46CSR
20
Mar 26, 2026
93.62CSR
20
Mar 26, 2026
95.02Turn 1 Score
18
May 22, 2026
8.21Overall Score (1-10)
12
Jun 8, 2026
91.28Accuracy
9
Jun 19, 2026
68.93Normalized Score
5
Mar 25, 2026