ResearchBenchmarksMathematical Reasoning on BeyondAIME (Mean@10)Mathematical Reasoning on BeyondAIME (Mean@10)Follow71.8Mean@10FlashMLA69.40870.02970.6571.271Feb 11, 2026Evaluation resultsMethodMethodLinksMean@10FlashMLABackbone=DeepSeek-V3.1...Backbone=DeepSeek-V3.1, Precision=BF162026.0271.8SnapMLABackbone=LongCat-Flash...Backbone=LongCat-Flash-thinking, Precision=FP82026.0270.2SnapMLABackbone=DeepSeek-V3.1...Backbone=DeepSeek-V3.1, Precision=FP82026.0269.9FlashMLABackbone=LongCat-Flash...Backbone=LongCat-Flash-thinking, Precision=BF162026.0269.5