ResearchBenchmarksInstruction Following on Muses-Bench Aligned n=187Follow93.16AccuracyDeepseek40.203253.951667.781.4484Jun 20, 2026Evaluation ResultsMethodMethodLinksAccuracyW.AccPer-User AccuracyDeepseekFramework=Muses-Bench,...Framework=Muses-Bench, Mode=Aligned2026.0693.1693.2494.07QwenFramework=Muses-Bench,...Framework=Muses-Bench, Mode=Aligned2026.0691.3691.8691.36GeminiFramework=Muses-Bench,...Framework=Muses-Bench, Mode=Aligned2026.0683.3583.3282.78GPTFramework=Muses-Bench,...Framework=Muses-Bench, Mode=Aligned2026.0642.2456.0754.55