Multi-task Language Understanding
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
99.7Accuracy
881
Apr 21, 2026
98.5MMLU Accuracy
456
Jun 16, 2026
94.7Accuracy
353
Apr 23, 2026
83.29Accuracy
143
Jun 23, 2026
86.4MMLU Score
116
May 27, 2026
90.46Normalized Accuracy
107
Jun 30, 2026
80.2MMLU Score
86
May 29, 2026
96.8Accuracy
66
Jun 19, 2026
82.05Accuracy
64
May 18, 2026
92.85Pass@1
64
Apr 14, 2026
95.2Accuracy
48
May 19, 2026
74.7Top-1 Accuracy
46
Jul 9, 2026
51.91Accuracy (%)
42
Jul 3, 2026
90F1 Score
41
Jun 4, 2026
76.88MMLU Pro Engineering Acc
41
May 19, 2026
95.71Accuracy
36
May 27, 2026
63.73MMLU Score
33
Jun 19, 2026
89.44Accuracy
33
Jun 19, 2026
92.8Physics Score
32
Jul 1, 2026
78.74Accuracy (5-shot)
31
Feb 26, 2026
91.21Accuracy
31
Jun 19, 2026
2.07Average Inference Time (s)
30
Mar 25, 2026
73.1Accuracy
27
Jun 29, 2026
28.79Accuracy
26
Mar 18, 2026
71.4Best Accuracy
25
May 11, 2026