Physix Frontier badgePhysix Frontier
PublishChinese | EN

Physix Frontier · News Briefing Card (QbitAI · Aug 20, 2026)

Moqi AI Unveils MoRA Architecture and KINO Robot

KEY FACTS

  • Moqi AI debuted at WRC 2026, launching its embodied model MoRA and the wheeled robot KINO.
  • KINO completed a 15-minute multi-zone household task on-site, covering cleaning, restocking, and laundry handling.
  • MoRA adopts an Agentic-Native approach, embedding goal maintenance and progress tracking into the low-level policy model.
  • Moqi has accumulated 30,000 hours of real-world data and plans to expand this to 150,000-200,000 hours within the year for model training.

KEY DATA

15 minutesDemo Duration
30,000 hoursCurrent Data Volume
150,000-200,000 hoursAnnual Data Target

PHYSIX OBSERVATION

Moqi aims to break the traditional hierarchical bottleneck where the 'brain commands the limbs' by granting native autonomy to the execution layer, thereby solving robustness issues in long-horizon tasks. This architectural innovation, paired with a focus on high-quality real-world data loops, signals a shift from flashy demos to actual operational efficiency. If this paradigm reduces reliance on high-level planning, it will significantly accelerate the deployment of general-purpose robots in complex, unstructured environments.

Source: QbitAI report