There is Zhang Xue in racing. There is WeRide in intelligent driving.
On May 17, Zhang Xue Motorcycle won its fifth title of the season at the Czech round of WSBK.
**△**No. 53 of the Zhang Xue team, which secured five consecutive victories at the WSBK Czech round
The title of “Five-Crown King” has been claimed by a Chinese motorcycle brand for the first time. It is an unprecedented achievement.
On the same day, in the final of the second China Intelligent Driving Competition’s Hefei round held in Hefei, the Chery Xingtu Exeed ES equipped with WeRide’s WRD 3.0 solution once again took first place with a score of 102.81. This was WeRide’s fifth consecutive championship in the competition.
Of course, being a Five-Crown King does not mean being invincible… WeRide had already run into a strong opponent in the qualifying round.
**△**In the Hefei qualifying round, both first and second place were WeRide solutions
But on closer inspection, that strong opponent was also WeRide’s own solution.
In two unconventional races, the same five consecutive victories were achieved. One rewrote the global hierarchy of top-tier two-wheeled racing; the other set a new Chinese record in four-wheeled intelligent-driving competition.
In both cases, the win did not come from marketing hype, but from hard-earned in-house technology development.
Zhang Xue’s MTC1000 electronic control system, bidirectional balance shaft, and variable valve technology all work around the patent barriers of Japanese manufacturers. Behind WeRide’s WRD3.0 lies the intensive refinement of the WeRide GENESIS virtual world model, plus feedback from a fleet of about 2,800 L4 vehicles. This is what truly brought L4 Robotaxi capability down to L2 mass-produced vehicles for the first time.
The two Five-Crown Kings born on the same day share astonishing similarities.
Zhang Xue Motorcycle replaced virtual simulation with real-world riding; WeRide ran through every track in advance through virtual simulation.
Their five consecutive victories are the result of the most practical self-improvement, the shortest feedback cycles, the highest test density, and the most thorough in-house control—compressing technology gaps that used to take 10 years into just 2 to 3 years, and breaking through in one stroke. WeRide founder and CEO Han Xu also posted about this on WeChat Moments, citing an allusion from Romance of the Three Kingdoms to describe the arduous journeys both companies have taken and the feat of “the few defeating the many.”
This China Intelligent Driving Competition is hosted by Diandong.com, and its lineup now includes many of the best-known names in autonomous driving.
**△**The in-house development camp of NIO, Li Auto, and XPeng, as well as solution providers such as Momenta, Horizon Robotics, Yuanrong Qixing, and Qingtzhou Zhixing, are also taking part
The course covers urban expressways, complex intersections, unprotected left turns, mixed traffic of people and vehicles, and more—in short, the exact road environment Chinese drivers face every day.
Starting from the Taizhou round in Zhejiang, WeRide went on to win five races in a row through Wenzhou, Jinhua, and Wuhu, before finishing in Hefei and setting a winning streak record in the competition.
In particular, in the Hefei qualifying round, the GAC Aion N60 equipped with WRD 3.0 made its debut. And it was running on just one Qualcomm Snapdragon SA8650P, whose computing performance is significantly lower than the dual Orin-X setup commonly used in the industry.
**△**WRD3.0 battling complex road conditions in the rain
As a result, the Chery Xingtu Exeed ES equipped with WRD3.0 took first place in qualifying.
And the debuting Aion N60 finished a very respectable second. It beat the third-place vehicle by as much as 24.95 points, decisively defeating the vehicle solution from a hugely popular heavyweight player. If the competition rules had not limited each player to only one vehicle in the final, WeRide might have ended up taking both first and second place. It would have been like turning the World Table Tennis Championships into a national games event.
The GAC Aion N60 and the Chery Xingtu Exeed ES (upgraded from OS2.8.0 to the latest one-stage end-to-end solution) both use the same WeRide WRD3.0 algorithm, yet deliver first- and second-place results on completely different compute platforms: dual Orin-X and a single Qualcomm 8650P.
This is not a one-off, chip-specific custom solution. It is a truly software-hardware decoupled, cross-platform, mass-production-grade intelligent driving system.
**△**WRD3.0 solution unveiled at the Beijing Auto Show
Coincidentally, Zhang Xue Motorcycle’s five-title run is also built on cross-platform capability. The 820RR-RS race bike and the soon-to-be mass-produced 820RR-R share the same engine and electronic control system, allowing technology proven on the track to be transferred directly into production vehicles.
One is two-wheeled racing; the other is four-wheeled intelligent driving. Both companies are proving the same point: truly robust technology does not collapse when the platform changes.
If Zhang Xue Motorcycle broke the long-standing dominance of top racing brands from Europe, Japan, and the US, then WeRide’s WRD 3.0, by winning five straight titles in China’s intelligent-driving competition, has also overturned the ingrained belief that “big brands and popular names are always stronger.”
How did WeRide achieve this?
Five consecutive victories did not happen by chance. WeRide’s technology foundation can be described as “walking on two legs” — one raises the floor, the other breaks through the ceiling.
The first leg is data feedback from nearly 2,800 L4 in-house autonomous vehicles.
This may sound familiar in the autonomous-driving industry—the “dimensionality reduction attack” that L4 players have talked about for years. But over all these years, only WeRide has truly made it happen.
WeRide operates about 1,300 Robotaxis worldwide. Together with its full lineup of L4 vehicles, including shuttle buses and sanitation vehicles, the total number reaches about 2,800. These vehicles run daily across more than 40 cities in 12 countries, continuously accumulating real-world road data under actual operating conditions. In particular, “long-tail scenarios” that are rarely encountered in ordinary road testing are extremely valuable.
**△**WeRide’s Robotaxi GXR driving in Riyadh, Saudi Arabia
That data is then fed directly back into WeRide’s in-house WeRide One general autonomous-driving platform, where the trained algorithms are applied to the WRD 3.0 L2++ solution.
As a result, WRD 3.0 is not starting from scratch. It is already a “veteran driver” that has logged tens of millions of kilometers in L4 scenarios. In other words, it is born with a high floor. From day one, it stands where other companies only reach after years of struggle.
The second leg is WeRide GENESIS, the in-house simulation world model.
In the virtual world, it can generate a wide variety of extreme scenarios at densities far beyond real roads: sudden cut-ins from adjacent lanes, unprotected left turns, pedestrians darting out, riders going the wrong way, and even natural disasters.
**△**Simulation scenes in WeRide GENESIS
GENESIS builds a complete closed loop of “scene generation → training → evaluation → diagnosis,” allowing the algorithm to rehearse these extreme situations again and again in a virtual environment before ever hitting the road.
That is what “raising the ceiling indefinitely” means. Situations that might only be encountered once in years of real-world driving and tens of thousands of kilometers can be repeated endlessly in WeRide GENESIS. You can even change conditions slightly and run variation drills.
For users who buy and daily use vehicles equipped with WRD 3.0, a high floor means the AI driver won’t make rookie mistakes in normal situations. In particular, under an end-to-end black-box solution, it avoids learning bad driving habits from human drivers.
A high ceiling means that when true hellish scenarios appear, it does not panic, and can navigate complex sections more rationally and steadily than a human driver.
**△**After being cut off, WRD 3.0 passed through the gate with two reverse maneuvers
WeRide founder and CEO Han Xu once said something interesting:
“The reason Wing Chun is strong is because Yip Man is fighting, not because Wing Chun as a martial art is inherently strong.”
In other words, the world-model technology paradigm has become popular, and nearly everyone in both L4 and L2 is following suit. In theory, any capable player should be able to achieve “mastery through one method” in both Robotaxi and mass-production L2.
But reality is different. At this stage, WeRide is the only company that has truly fed tens of millions of kilometers of L4 fleet data into its L2 solution and thoroughly trained every extreme scenario through its own simulation system.
Behind that are the effort and sweat that other players have not yet paid. Slimming down from L4 to L2+ is like a marathon runner training explosive power for the 1,500-meter race. It is not just about trimming a little endurance from long-distance capability.
**△**WeRide’s in-house 2000 TOPS compute platform, HPC3.0
From a technical standpoint, this requires not only basic knowledge distillation, but also fine-grained know-how down to the code level, including Mixed-Precision quantization before distillation, model pruning, dynamic inference optimization, and MoE architecture optimization…
Looked at from another angle, this technical accumulation explains why players who “upgraded dimensions” from L2 cannot truly be called real L4s.
The same is true for Zhang Xue Motorcycle. Its in-house MTC1000 electronic control system, lightweight frame, and integrated aerodynamic design were all not off-the-shelf adaptations, but were built one lap at a time on the track.
What the two companies share is that they firmly hold the initiative in R&D and have used the most down-to-earth, and therefore most honest, methods to drill through technical barriers one by one.
How should we understand these two “Five-Crown Kings”?
Looking at the core technologies behind these two Five-Crown Kings side by side reveals a fascinating resonance.
First, their technical logic is fundamentally the same.
Zhang Xue Motorcycle’s X3 engine and MTC1000 electronic control system, and WeRide’s WeRide GENESIS simulation world model plus L4 data feedback, both essentially create a “closed loop.” In other words, data generated in actual operation is quickly fed back into development, enabling ultra-high-frequency iteration.
Zhang Xue Motorcycle sends data back immediately after each lap on the track and can upgrade to a new version within 24 hours. WeRide runs tens of thousands of extreme scenarios in the virtual world and updates its algorithm models overnight.
Both companies have compressed the “development-test-optimization” cycle from years to months, and from months to days.
Second, they share the same goal: breaking monopolies. Zhang Xue Motorcycle directly challenged giants like Ducati, Yamaha, and Honda, pushing them off the podium in WSBK.
WeRide’s WRD 3.0, in the China Intelligent Driving Competition, directly defeated what is regarded by ordinary users as almost the “ceiling” of the field.
The meaning of five consecutive titles is not the trophy itself. It lies in proving that China-made L2++ solutions—especially the mass-production solutions of L4 players—can fully outperform top-tier manufacturers in real user commuting and travel scenarios.
This also leads to a more practical conclusion. With Tesla FSD accelerating its entry into China, reports have also recently surfaced about intelligent-driving-related hiring in the domestic market.
And in terms of both technical system and business model, the only Chinese company that may truly be able to compete with FSD is WeRide.
**△**GAC Aion N60 equipped with WRD3.0
Because Elon Musk will not get involved in arguments between the L2 and L4 camps, nor will he care which technical route is more “orthodox.”
That means the traditional metrics the L4 camp has competed on—such as the number of cities opened or the scope of operations—have no direct counterpart for Tesla. Likewise, when L2 players boast about deployment counts or disengagements per 100 kilometers, Elon Musk simply does not care. Guided by the first principles of autonomous driving, Tesla is always developing toward full-scenario, full-time, full-region capability—
just like WeRide, which has always preferred action over talk.
But beyond technology, WeRide also has an advantage over FSD in another way.
Its relative advantage is that WeRide has already accumulated a massive amount of real-world scene data in China and has walked the road to market before FSD did.
Its absolute advantage is that WeRide has already deployed WRD 3.0 on the Aion N60, which has a starting price of 106,800 yuan, and offers it as standard equipment. The cost/profit is included in the vehicle price. By contrast, FSD’s one-time purchase price is 64,000 yuan.
**△**WRD3.0 passing each other in a narrow road
When advanced intelligent driving is no longer a “premium option” but standard equipment across all trims, the rules of the game have already been rewritten.
What is even more noteworthy is that WeRide’s five-title-winning capability has already begun to be replicated at scale in production models.
WeRide has already secured adoption wins across about 30 vehicle models, covering three brands under GAC Group—Aion, Hyptec, and Trumpchi—and five brands under Chery Group—Exeed, Tiggo, Lepas, Omoda, and JAECOO. Its market footprint now spans China, Europe, Southeast Asia, and the Middle East.
Among them, the Aion N60, WeRide and GAC Group’s first production vehicle together, was launched in April this year. Within two weeks of pre-orders opening, it had received over 8,300 orders, and by early May that number had surpassed 9,000. The car has shown remarkable performance even in “hellish scenarios” such as Guangzhou’s urban villages and evening rush hour, earning praise from automotive reviewers on Bilibili as the only zero-intervention vehicle on the “intelligent-driving Nürburgring”—and its rivals were cars in the 200,000-yuan or even 300,000-yuan class.
So let’s return to the original question: was it a coincidence that two Five-Crown Kings emerged on the same day?
It was both a coincidence and not a coincidence. It was a coincidence that they happened on the same day. It was not a coincidence that they share the same underlying logic: the shortest feedback cycles, the highest test density, and the most thorough in-house control—using 2 to 3 years to break through technology barriers that once took 10 years.
Zhang Xue Motorcycle turned real track combat into its development process, while WeRide ran every course ahead of time in virtual simulation.
**△**Inside WeRide GXR Robotaxi
Now that multimodal large models plus data-driven development have become the industry consensus in autonomous driving, the “L4 vs. L2 route war” no longer really matters.
What truly matters now is just one thing: across different cities, road conditions, and vehicle models, who can consistently enable the AI driver to perform at a high level.
As Han Xu said, don’t look at the style—look at the result: “Wing Chun is strong because Yip Man is fighting.”
Through five consecutive victories, WeRide has proven that first. Yip Man is already in the ring.