ResearchBenchmarksNext Step Generation on PeanoBench 21 incorrect proofsFollow3.7AccuracyLeanTutor2.7643.0073.253.493Jun 10, 2025Jul 18, 2025Aug 25, 2025Oct 2, 2025Nov 9, 2025Dec 17, 2025Jan 24, 2026Evaluation ResultsMethodMethodLinksAccuracyRelevanceReadabilityAnswer LeakageLeanTutorFeedback Type=Next StepFeedback Type=Next Step2026.013.73.94.41.1LeanTutorFeedback Type=Next StepFeedback Type=Next Step2025.063.73.94.41.1BaselineFeedback Type=Next StepFeedback Type=Next Step2026.012.82.94.62.2BaselineFeedback Type=Next StepFeedback Type=Next Step2025.062.82.94.62.2