OlympiadBench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
OlympiadBench
83.82Pass@8 Accuracy
5
OlympiadBench
57Accuracy
5
OlympiadBench
74.5Avg@4 Score
4
OlympiadBench AIME24 AIME25 Minerva Average
27.7Average Score
4
OlympiadBench
45.2Score
4
OlympiadBench (val)
54.16Accuracy
4
OlympiadBench (test)
108Efficiency Ratio
4
OlympiadBench
25,517Tokens per Correct
4
OlympiadBench
11.9Proof-level Accuracy
3
OlympiadBench (first 40 problems)
10Time (minutes)
3
OlympiadBench
53.66Best-of-64 Accuracy
3
OlympiadBench Math
50.85Accuracy
2
OlympiadBench
43.52Pass@50
2