Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Mathematics on OlympiadBench (pass@1 %)

78.72Pass@1 Accuracy

COSE

9.955227.807645.6663.5124Apr 21, 2025Jun 26, 2025Sep 1, 2025Nov 7, 2025Jan 13, 2026Mar 21, 2026May 27, 2026
Updated 1d ago

Evaluation Results

MethodLinks
2026.05
78.72
2026.05
72
2026.05
69.4
2026.05
68.9
2026.05
68.6
2026.05
66.8
2026.03
57
2026.03
56.45
2026.03
56.24
2026.03
56.2
2026.03
53.38
2026.03
52.5
2026.03
52.25
2026.03
52.07
2026.03
52.01
2026.03
51.8
2025.10
50.7
2025.10
50.2
2026.05
47.6
2026.05
46.3
2025.10
45.8
2026.05
44.2
2025.10
43.3
2025.10
41.7
2025.10
41.3
2025.10
41
2025.10
40.8
2025.10
40.4
2026.05
38.1
2025.10
37.2
2025.10
36.6
2025.10
34.8
2026.03
34.8
2026.03
34.8
2025.10
34.4
2026.03
34.25
2026.03
34.2
2025.10
20.6
2025.10
20.3
2025.04
19.05
2025.04
18.72
2025.04
18.55
2025.04
18.43
2025.04
18.37
2025.04
18.31
2025.10
18.3
2025.10
18.2
2025.04
17.6
2025.04
16.6
2025.04
16.08
2025.10
12.6