Tesla’s Full Self-Driving (Supervised) version 13.2.6 has shown that there is no area that new advances in end-to-end AI can’t tackle, and there is no better testing ground than the busy streets of Guangzhou’s roads. From driving through narrow lanes to dealing with assertive lane changing and shared roads with scooters, it seems like Tesla’s latest FSD update has learned how to read China’s complexity like a human.
FSD Meets the Urban Jungle
In recent tests and actual video recordings, the FSD v13.2.6 was observed to handle turns in the city of Guangzhou’s intersections and overpasses in a rather professional manner. Where the previous versions proved to be tentative in high-pressure driving dynamics environments, this update appears significantly more logical and prompt in decision-making.
There is little doubt in its actions; even a roundabout is approached cautiously but assertively, and the same goes for passing by another car that has parked alongside the road and is occupying more than one lane.
Guangzhou's streets are legendary for their crazy mix of cars, bikes, and everything in between 🔥
We put Tesla's FSD to the ultimate test: navigating this urban jungle 😱
Did it crash? Did it freeze? Or did it actually… work? Hit play and see the real story. This is the FSD… pic.twitter.com/yDEhkZKu5a
— TCMeslanaire (@TCMeslanaire) April 20, 2025
Still more impressive is how it “understands” such driving etiquette, which is more appropriately described as unspoken rules and regulations guiding the use of roads. While younger drivers or those brought up in the United States may expect the roads to be neat and all to adhere to the rules while driving, the Chinese use roads in a more aggressive as well as flexible manner. FSD version 13.2.6 seems to have caught on.
More Human Than Human (Driver)—in China
Looking at actions that Tesla’s AI takes on the road, it can often be assumed that it is more Chinese than American, even though the company is based in California. As for the users from China who tried FSD, they have reported that the system is much less stiff and seems to act almost like a human.
For example, it looks for scooters at the intersection and manages to navigate shared lanes, both of which are features of the local version of the system but not the American one; it accelerates at the time when it is safe to do so during a red light, which is another feature typical of the local driving manners but not included in the American version of the software.
Such regional flexibility suggests shifting towards North American training data and Tesla’s overall end-point neural network learning initiative. Rather than having pre-programmed rules, the AI builds up driving behavior from actual experience as observed in the territory, making it quite realistic for each country respectively.
Off-Road Capability
This is even more impressive since FSD can smoothly work on more rugged roads—an environment not exactly created for it. Regardless of being on gravel roads in remote areas of Guangdong or unpaved backstreets near industrial districts, version 13.2.6 still has good lane positioning, speed management, and obstacle avoidance, even if roads are not marked and the GPS signal is fluctuating. This is made possible by pure vision and neural networks of Tesla and gives a clear demonstration of the strength of its end-to-end artificial intelligence.
It’s far from being able to go plunging off the beaten track, but it does represent a step forward in the capability of FSD being restricted to well-maintained city roads.
Model Y Dominates China’s NEV Market in March
All of this comes as Tesla enjoys growth in stardom in China due to the Model Y. In March, it sold as many as 48,189 Model Y EVs in China alone, and the company easily holds the largest market share of new energy vehicles in the country. Therefore, Tesla’s affordable, extensive, and technologically equipped model, FSD, can be considered a perfect fit for consumers in China who are craving both luxury and functionality.
The success of the Model Y also proves that Tesla has shifted up the trust of consumers in its technology within one of the most hotshot EV markets in the world. BYD and NIO have raised competitive pressure on Tesla in every aspect where AI has not been the key way of gaining an edge.
