ResearchBenchmarksMath Reasoning on HMMT Feb 25 (pass@8, mean@8)Follow18.5Mean @8MOPD4.52248.151211.7815.4088May 12, 2026Evaluation ResultsMethodMethodLinksMean @8Pass@8MOPDBase Model=Qwen3-4BBase Model=Qwen3-4B2026.0518.519.58GRPOBase Model=Qwen3-4BBase Model=Qwen3-4B2026.0516.9220.94SDPOBase Model=Qwen3-4BBase Model=Qwen3-4B2026.058.5913.61Qwen3-4Bstatus=base modelstatus=base model2026.055.0610