Humanity's Last Exam (HLE)
Benchmarks
TaskDatasetSOTA resultTrendResults
Humanity's Last Exam (HLE) (test)
45.8Accuracy
14
Humanity's Last Exam (HLE) text-only
36.6sHLE Score
11
Humanity's Last Exam (HLE) (test)
72.19Accuracy
10
Humanity's Last Exam (HLE) Medical
20.8Accuracy
7