ResearchBenchmarksModel Ranking Correlation Analysis on MathVistaFollow0.92Spearman Correlation (10% subset)M3IRT0.56640.65820.750.8418Mar 3, 2026Evaluation ResultsMethodMethodLinksSpearman Correlation (10% subset)Spearman Correlation (20% subset)Spearman Correlation (30% subset)Spearman Correlation (40% subset)Spearman Correlation (50% subset)M3IRT2026.030.920.940.930.930.93FlashEval2026.030.890.910.910.910.91Random2026.030.890.910.920.930.93M2IRT2026.030.810.840.880.910.93IRT2026.030.810.880.910.930.94TinyBenchmarks2026.030.790.840.860.880.88MIRT2026.030.580.670.720.760.81