Motoniq robot agent hits 92% on perturbed LIBERO tasks, next best 53%
hbouammar · x · 2026-10-24
Motoniq Labs reports its robot agent achieves 92% task success across ten perturbed LIBERO tasks with new layouts, object positions and drawer locations — the next best agent tested managed only 53%. Other baselines: ASPIRE 51%, CaP-Agent0 12%, Claude-based agent 9%.
More from Embodied
- Samsung foldables at 3% after 7 years; blogger predicts Apple's foldable iPhone hits ~20% by 2028 — mitchdeg · 2026-10-25
- Dev feeds Grok thermal camera and live video to auto-debug overheating hardware — Baconbrix · 2026-10-25
- What a 4GB model can do on a robot: Nemotron Bolt runs locally on Reachy Mini — lmoroney · 2026-10-25
- What can a 4GB model do on a robot? Nemotron 3.5 Bolt E4B tested on Jetson — lmoroney · 2026-10-25
- Nebius launches open-source Physical AI Workbench with NVIDIA, cutting robotics engineering time by 40% — lukas_m_ziegler · 2026-10-25
- Runway co-CEO Anastasis Germanidis to headline State of AI NYC meetup on Oct 27 — nathanbenaich · 2026-10-25