LLM Alignment
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
87.08GPT-4o Win Rate
45
Jun 1, 2026
8.34AlpacaEval 2 WR
27
Mar 4, 2026
87.85WR
27
May 19, 2026
61.52LC Win Rate
25
May 19, 2026
86Win Rate
24
Jun 11, 2026
51.4Win Rate
24
Apr 21, 2026
25.24Win Rate
24
May 11, 2026
0.87Diversity
23
Apr 28, 2026
-0.02Average Oracle Reward
20
Jun 4, 2026
152.56TFLOPs
20
May 25, 2026
21AlpacaEval 2 Win Rate (WR)
18
Mar 4, 2026
69.8Win/Tie Rate vs Vanilla (GPT-4o)
16
Mar 4, 2026
89.3Diversity
15
Feb 26, 2026
1.53Win Score
13
Mar 4, 2026
31.58LC WR (%)
12
May 25, 2026
39.72LC WR
12
May 25, 2026
40.9LC WR
12
May 25, 2026
8.78LC Win Rate
12
May 11, 2026
94.33Win Rate
12
Mar 24, 2026
49BWR (Median)
12
Mar 16, 2026
97.1Win Rate
12
Feb 26, 2026
54.38Average Percentage Score
9
Feb 26, 2026
28.34Win Rate
8
Jun 11, 2026
57.53GPT-4o Win Rate
8
Jun 1, 2026
80.6Win Rate (KL, alpha=1)
8
Feb 26, 2026