Real-world Understanding
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
80.6Win Rate
25
Apr 8, 2026
64Score
20
Feb 27, 2026
55.5Distance Error
19
Feb 27, 2026
85.6Accuracy
18
Mar 12, 2026
71.6Score
11
Mar 12, 2026
86.9Score
10
Mar 12, 2026
70.07Score
9
Mar 5, 2026
73.8Accuracy
8
Mar 12, 2026
77.9Accuracy
8
Mar 12, 2026
42.5Accuracy
7
Mar 12, 2026
68.2RealWorld QA Score
5
Feb 26, 2026