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

Mathematical Problem Solving on AIME 2025 (Accuracy, Time, and Token Usage)

86.7Accuracy

RecursiveMAS

15.25233.80152.3570.899Apr 28, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.04
86.78,9815,342
2026.04
868,1785,314
2026.04
807,7846,338
2026.04
73.319,30423,651
2026.04
71.38,4629,397
2026.04
70.714,38016,372
2026.04
342,7271,586
2026.04
33.32,3671,614
2026.04
30.71,8291,622
2026.04
242,3802,993
2026.04
23.34,2475,318
2026.04
186,1838,645