Physix Frontier · News Briefing Card (QbitAI · Aug 21, 2026)

Qianxun AI Hits Near-100% Success in Long-Horizon Tasks

KEY FACTS

Qianxun AI re-enacted its WAIC 'Living Room Cleanup' demo at the World Robot Conference to showcase rapid system evolution.
Autonomous reasoning success rates for tasks like placing a Coke in the fridge or loading dishes into the dishwasher rose from 80%-90% to over 99%.
Time required for navigation target confirmation dropped by nearly 50%, while execution success rates for generalized tasks such as picking up trash reached 85%.
The company introduced the concept of 'time density,' leveraging a full-stack technology loop to achieve significant performance leaps within 30 days.

KEY DATA

80%-90% to 99%+Core Task Success Rate Improvement

Nearly 50%Navigation Time Reduction

Down to 1/10 of traditional methodsData Collection Cost Reduction

From 30% to 95%Data Usability Increase

PHYSIX OBSERVATION

Competition in embodied AI has shifted from flashy demonstrations to engineering efficiency and deployment speed. By re-staging an old demo and quantifying progress, Qianxun proves it has built an automated loop from data collection to model iteration. This rapid evolution capability, centered on 'time density,' is more critical than any single demonstration, signaling that the industry is moving past the lab stage toward commercialization with robustness in real-world scenarios.

Source: QbitAI report