ResearchBenchmarksStructured video reasoning on Keys and Doors OverallFollow99.2Exact Match PercentageTBS-2050.42463.08775.7588.413Jun 11, 2026Evaluation ResultsMethodMethodLinksExact Match PercentageTime (ID)Time (OOD)TBS-20Budget=20Budget=202026.0699.22.618.13BoN-20Budget=20Budget=202026.0697.72.8619.66TBS-10Budget=10Budget=102026.0696.52.599.31BoN-10Budget=10Budget=102026.0694.52.519.89BoN-5Budget=5Budget=52026.0689.22.14.95BoN-1Budget=1Budget=12026.0652.311