Tesla FSD in China: Real-World Tests Against Top Rivals
Tesla’s “FSD” in China has attracted intense interest because engineers trained it on publicly available footage of complex Chinese roads. A recent CarNewsChina study compares the “FSD” system with advanced driver assistance systems from Xpeng, Xiaomi, Li Auto, Smart and Volkswagen to gauge real-world autonomous driving performance across busy urban routes.
Tesla markets “FSD”, short for “Full Self-Driving”, as an advanced driver assistance system that still demands continuous driver supervision. In China, it appears as “FSD智能辅助驾驶软件”, which translates to “FSD Intelligent Assisted Driving Software”. Chinese bloggers evaluated performance by testing the cars under identical road conditions and sharing the results widely on social platforms to facilitate like-for-like ADAS comparisons.
In a live AutoLab demonstration, testers pitted a 2024 Tesla Model Y with HW 4.0 V13.2.6 against an Xpeng P7+ Long Range Max running XOS 5.5.4.8901. Both cars approached a challenging roundabout and took the first exit cleanly. In a follow-up run, the Model Y delayed a lane change, and the driver intervened, so the car could not complete the manoeuvre because a vehicle on the left blocked the intended path.
Another AutoLab session on Weibo lasted 1 hour and 21 minutes, comparing the Model Y’s FSD V13 with Xiaomi’s SU7 on HyperOS 1.5.5. The team aimed to follow identical routes, although slight deviations crept in. The Xiaomi car demonstrated notable system stability and maintained a higher average speed, whereas the Tesla required six driver interventions triggered by unexpected lane changes and emergency stops.
AutoLab also conducted its longest trial for more than three hours, directly comparing the Model Y V13 with the Li Auto L9 AD Max V13. Both cars showed issues typical of today’s driver assistance technology. Testers recorded 16 driver interventions in the Tesla versus 10 in the Li Auto, which signpost areas where each brand can refine perception and planning.
In a separate 90-minute evaluation, the Smart #5 with Smart OS 2.1.1 CN was tested against the Tesla Model Y V13. The Smart needed fewer driver interventions than the Tesla, an outcome that again indicates scope for improvement within Tesla’s current software stack when handling dense traffic and frequent junctions.
Despite these setbacks, Tesla’s “FSD” shows clear potential in China’s dynamic driving environment and competes closely with domestic ADAS solutions from leading brands. The system navigates complex traffic scenarios and roundabouts with increasing confidence, although it still requires timely driver input to ensure optimal performance. As training data expands and algorithms mature, the technology appears poised to enhance its consistency and reduce the need for interventions on Chinese roads.
