Mystery agent method tops RoboDojo at 73.33, first past 70 points and 12 points ahead of runner-up
jiqizhixin · x · 2026-10-30
The RoboDojo-Sim leaderboard (66 models) has a new #1: an unidentified agent-based method scored 73.33 — the first past the 70-point barrier — with a 68% success rate, beating the runner-up by 12 percentage points. It reportedly outperforms teleoperation in precision manipulation, bimanual coordination, and motion smoothness. The approach is neither a VLA nor a world model; the chart labels the top entry as agent model Q0-Preview, with GO-3 Nano Preview and PyRUA-Lean further down.
More from Embodied
- Cute wheeled desktop robot serves drinks at LongJourney VC demo night — aniketmaurya · 2026-10-30
- Tiny M5Stack robot fangirls over Teknium in English and Japanese via LFM3 Audio — alexcovo_eth · 2026-10-30
- Artificial skin lets robot fingers grasp pencil lead and dodge needles, reading 10,000+ sensors 100x/sec — CurieuxExplorer · 2026-10-30
- Dyna Taku design praised as sharing 1X NEO's instant-appeal quality — CyberRobooo · 2026-10-30
- Honeywell bets on physical AI, targets $1 billion India revenue in 2-3 years — nordicinst · 2026-10-30
- BindCraft protein design on an M5 Max MacBook rivals RTX 5090, benchmarks show — AllThingsApx · 2026-10-30