ResearchBenchmarksMathematics on MATH GSM8K (Competition Split)Follow3.6ECEMARGIN2.728.6614.620.54May 21, 2026Evaluation ResultsMethodMethodLinksECEMARGINPhase=Phase 2Phase=Phase 22026.053.6HistPhase=Phase 2Phase=Phase 22026.0515.4PlattPhase=Phase 2Phase=Phase 22026.0515.7TempPhase=Phase 2Phase=Phase 22026.0516.9RawPhase=Phase 2Phase=Phase 22026.0525.6