ResearchBenchmarksDirect conversation on Conversational User Study DirectDirect conversation on Conversational User Study DirectFollow10Rank 1 CountFrontier-0.42.357.7Nov 10, 2025Evaluation resultsMethodMethodLinksRank 1 CountRank 2 CountRank 3 CountFrontierTask=DirectTask=Direct2025.111071ConvFillTask=DirectTask=Direct2025.11882Base SLMTask=DirectTask=Direct2025.110315