ResearchBenchmarks3D Layout Generation on User Study (test)Follow1,905Elo RatingR3L1,250.841,420.671,590.51,760.33May 7, 2026Evaluation ResultsMethodMethodLinksElo RatingWin Rate (vs LayoutGPT)Win Rate (vs Holodeck)Win Rate (vs LayoutVLM)Win Rate (vs R3L)R3L2026.051,905252625—LayoutGPT2026.051,511—21211LayoutVLM2026.051,308415—2Holodeck2026.051,2766—111