Tesla beats NYC‑LA FSD record

Tesla’s latest self-driving flashpoint is not a robotaxi launch or a regulator ruling. It’s a road trip — a very long one. Dan Burkland says he drove a Tesla Model 3 from Manhattan to Redondo Beach in 49 hours, 55 minutes, and 57 seconds with Full Self-Driving engaged the whole way and no driver disengagements. Tesla then amplified the post, which is why this turned from niche car-forum bragging rights into a real debate about what the company’s software can actually do. ### What actually happened? Burkland’s run started at the Red Ball Garage in Manhattan and ended at the Portofino Hotel in Redondo Beach — the classic Cannonball route. He posted that the car used FSD v14.3.2 for the entire trip and that the human never took over steering, braking, or acceleration. The claimed distance was about 2,833 miles. ### Why are people calling it a record? (evxl.co) Because there was already a smaller subculture keeping score. In January 2026, Alex Roy and collaborators said they completed the first zero-intervention Tesla FSD Cannonball from Los Angeles to New York. Burkland’s east-to-west run is being framed as the new zero-intervention benchmark because the posted time is about 8 hours faster than that January mark. ### Does “zero interventions” mean autonomous? No — and this is the part that gets muddy fast. Tesla’s own product page says Full Self-Driving is “Supervised,” requires active driver supervision, and “does not make the vehicle autonomous.” The owner’s manual is even blunter: the driver must stay attentive and be ready to take over at all times. So a no-disengagement trip is impressive as a software endurance test, but it is not the same thing as an unsupervised robotaxi trip. (teslarati.com) ### Why is highway driving the easier version? Because long interstate stretches are the friendliest environment for driver-assist systems. Lanes are clearer. Traffic flows in the same direction. There are fewer weird edge cases than on urban streets with pedestrians, bikes, construction, and ambiguous intersections. A coast-to-coast run still includes merges, weather changes, fuel stops, and city entry points, but the bulk of it is exactly the kind of structured driving modern ADAS handles best. (tesla.com) That makes this a meaningful milestone — but also a selective one. ### What changed with v14.3.2? Tesla’s recent v14.3.2 release notes point to a bigger architectural shift than a normal point update. Tesla-focused release tracking says the software unified models across FSD, Actually Smart Summon, and Robotaxi behavior. If that description is accurate, this run is notable partly because it was done on a version Tesla seems to view as a bridge between its consumer supervised stack and its future driverless stack. (tahaabbasi.com) ### So why are safety people still skeptical? Because one clean demo does not settle the safety case. NHTSA opened a defect investigation into FSD behavior tied to potential traffic-law violations, and in March 2026 the agency escalated scrutiny around reduced-visibility performance. Basically, the question regulators care about is not whether a Tesla can make one heroic trip. It’s whether the system behaves safely and predictably across millions of messy, ordinary miles. (notateslaapp.com) ### Does Tesla have broader safety data? Tesla does publish an FSD Supervised safety page and says collision likelihood drops when the system is engaged under active supervision. But even Tesla separates that claim from autonomy. The company still labels the feature supervised, still requires driver attention, and still treats the human as the fallback. That distinction matters more than the viral clip. (static.nhtsa.gov) ### Bottom line? This run looks like a real milestone for Tesla’s driver-assistance software — especially as a long-haul consistency test. But the catch is simple: a coast-to-coast zero-disengagement record is evidence of progress, not proof that Tesla has solved self-driving. (evxl.co) (tesla.com)