The meaningful part of XPENG’s latest robotaxi move is the invitation code. By opening online registration for XPENG YOYO, the automaker has begun taking a bounded ride-service test beyond its own employees. For a prospective passenger, that changes the nature of the exercise: the test now has to make sense to someone outside the company, while still keeping access controlled. The question is no longer only whether an autonomous system can move a car. It is whether the booking, pickup, payment, and drop-off experience can hold together when the rider is not part of XPENG.
XPENG named the service and opened the registration channel on October 8. The XPENG newsroom describes a progression from internal workflow work to employee testing and now to an invitation-based external trial. TechNode likewise identifies the move as a step toward public trials. What matters for readers is the restraint built into the current step: an invitation code can expose a service to people outside the company without presenting it as an unrestricted robotaxi launch.
XPENG is testing more than a vehicle feature. It is trying to connect an electric vehicle, autonomous-driving software, AI, and a passenger-facing service into one operating offer. That is a different commercial problem from putting an autonomous system through an internal run. A ride service has to handle the moments around driving as well as the drive itself, and the invitation process makes those handoffs part of what XPENG can now examine.
Invitation Codes Put the Ride Workflow Under Pressure
The company says its booking, pickup, and drop-off workflow was tested internally before the employee phase. It also says it charged RMB 1 per ride during earlier internal testing, using the payment step to validate the service workflow. Those choices explain why the invitation code matters. The company is not simply widening access to a vehicle; it is carrying a tested sequence into a more realistic, but still bounded, setting.
That distinction sets the useful standard for reading the trial. An external registration channel marks controlled external-user testing beyond XPENG’s staff. It is not evidence of a broad rollout, commercial-scale operation, or generally driverless performance. The bounded format lets the company test the practical seams of a ride service without claiming that every operational question has been settled. For a reader deciding what has changed, the answer is specific: an internal process has become an external-user test, not a citywide transportation service.
The service also gives a more concrete frame to XPENG’s robotaxi effort than technical labels alone can provide. A passenger does not experience onboard compute as an abstract specification. The passenger encounters a sequence: booking, being picked up, completing payment, and reaching a destination. The earlier internal workflow gives XPENG a starting point, while invitations create a way to see whether that sequence remains workable outside the company’s own staff setting.
The distinction is practical: a rider needs the service as a complete customer interaction. Readers can find related background in EastFrontier’s earlier XPENG robotaxi coverage. This controlled test begins to probe that complete interaction.
XPENG’s Technical Stack Meets a Narrow Guangzhou Test
XPENG says a mass-produced robotaxi rolled off its Guangzhou line in May. It also says it obtained Guangzhou remote-testing qualification in August for designated road classes without a safety operator in the driver’s seat. The qualification is important because it defines a path for testing without putting a safety operator in that seat, but it is not blanket approval to provide passenger service under any condition. The operating boundary is part of the story, not a footnote to it.
The company describes the vehicle system as using four in-house Turing chips and 3,000 TOPS of effective onboard compute, with VLA 2.0 and a vision-first approach that does not use LiDAR or HD maps. These are XPENG’s technical specifications, not an independent measure of performance. They establish the system the company intends to put through the trial, but specifications do not settle whether the entire service will be reliable in external use. The invitation program is where the vehicle and service layer begin to meet a practical test.
A vehicle coming off a production line and a fleet delivering rides are different milestones. XPENG’s Guangzhou production claim is a separate company-reported milestone alongside the invitation trial. The registration step asks whether that premise can become a bounded service. Keeping those milestones separate prevents a familiar mistake in autonomous-driving coverage: treating a production claim or a test qualification as proof that a public transportation operation is already in place.
There is also a wider strategic choice in XPENG’s structure. The company says it will provide vehicles, autonomous-driving software, and AI, while local partners operate fleets. That arrangement puts the operating model alongside the driving system. Readers can find related background in EastFrontier’s earlier XPENG technology-stack coverage. The current trial makes the operating question tangible: who can turn a company-supplied system into a dependable ride service?
Local Fleet Partners Become the Decisive Operating Test
The partner model makes YOYO’s invitation-only phase a test of coordination. XPENG’s role, as the company describes it, is to supply the core vehicle and technology; local partners are intended to run fleets. That division turns the next stage into a test of coordination. The service must connect the company’s vehicle, autonomous-driving software, and AI with the fleet operator responsible for putting the ride offering into operation.
XPENG has described expansion in 2027, a dedicated model, and break-even as plans or targets, rather than present outcomes. Those ambitions may define the company’s direction, but they do not turn the invitation trial into evidence of either profitable operation or a scaled service. The current development is valuable because it narrows the distance between internal validation and a service people outside XPENG can attempt to access. It does not eliminate the distance.
That is a practical standard for assessing what comes next. A robotaxi program can accumulate impressive technical descriptions and still leave the rider-facing operation untested. Conversely, an invitation code is not a substitute for evidence about where a service runs, how many vehicles support it, whether it is reliable, how it performs on safety, or which partners operate it. Those are the details that show whether a controlled trial has become a durable service.
The genuine next test is disclosed external operations: the service area, fleet, reliability, safety performance, and named partner arrangements.