Tesla’s Cybercab now has 8 crashes on file with NHTSA, barely two months after the purpose-built robotaxi started giving public rides in Austin.

Four happened in driverless “Robotaxi” service, and none topped 5 mph. The other four happened with a test driver on board and Tesla’s driver-assist engaged, including one where a Cybercab hit a car at 53 mph on a highway in Philadelphia.

The data points at what we think is the biggest limitation of Tesla’s self-driving program right now: speed.

Tesla files Cybercab crashes under NHTSA’s Standing General Order in two separate datasets.

Industry observers note that the first is the ADS (Automated Driving System) file, which covers Level 4 driverless operation. That’s the Robotaxi service, where nobody is in the driver’s seat. NHTSA’s latest ADS release, published this month, contains 43 unique Tesla reports, up from 22 in July. Four are Cybercabs:.

According to recent reports, those are minor. Two weren’t Tesla’s fault at all, and none involved an injury.

In a recent development, the second bucket is the Level 2 ADAS file, which covers driver-assist systems like Autopilot and “Full Self-Driving” (Supervised). Tesla has been testing Cybercabs fitted with steering wheels and a human behind them, often in places where it has no Robotaxi permit. Those crashes land in the ADAS file, with the driver type coded “In-Vehicle (Commercial / Test)” and the system “Verified Engaged.”.

There are four of them, all in September 2026:.

Tesla redacted the narrative of all four as confidential business information, so there’s no way to tell from the public data who was at fault or what the system was doing. Pennsylvania State Police, Austin Police, Lehi Police, and Florida Highway Patrol are listed as the investigating agencies.

In a recent development, the Philadelphia crash is not the Broad Street crash where a Cybercab lost its door panel that we reported in October. That one happened on a city street in October. This one happened on a highway in September, at around 10 p.m. local time.

Tesla has no permit to run automated vehicles in Pennsylvania, and it doesn’t run Robotaxi in Utah either. In Utah, a Cybercab was photographed in Park City in September with a driver visible behind the wheel.

Two of the eight Cybercab crashes involved injuries. Both are in the Level 2 file.

Recent findings suggest that across all 43 Tesla Robotaxi reports in the ADS file, the highest pre-crash speed is 29 mph. That was a driverless Model Y with two riders hitting a pothole in Dallas in September, after which one rider said they sought medical care. The median speed across all 43 reports is 1 mph.

In a recent development, not one of those 43 crashes happened on a highway. Tesla coded 21 on streets, 15 in parking lots, and 7 at intersections.

It has been revealed that those numbers say more about where Tesla lets the system drive than about the software itself. As we reported last month, Robotaxi sticks to low-speed routes inside small geofenced zones and largely stays off highways.

The Robotaxi Safety Report Tesla published on October 21 claims zero people injured by Robotaxi over 3,112,541 driverless miles between January 22 and October 18. It doesn’t mention the Cybercab’s test fleet, and it excludes every mile driven with a safety monitor on board.

Take the speed cap off and the picture changes.

In a recent development, in the Level 2 data, Teslas with Autopilot or FSD engaged regularly crash at highway speeds. Our master dataset of Tesla’s ADAS reports holds 69 fatal crashes. Among the 52 where Tesla recorded a speed, the median was 53 mph, the same speed the Cybercab was going in Philadelphia, and 29 happened at 50 mph or more.

Today alone, we matched three fatal crashes to Tesla’s latest filings:.

It has been revealed that engagement within 30 seconds of a crash, which is what triggers these reports, doesn’t prove the system caused it. But the pattern is consistent. At low speeds in a tight geofence, Tesla’s crashes are scrapes with chains, curbs, and potholes. At highway speeds, they can kill.

Sources indicate that the Cybercab test fleet is where those two worlds overlap. It’s the same vehicle Tesla runs driverless in Austin, and the one Cybercab crash on record at real speed happened at 53 mph on a Philadelphia highway.

The Cybercab’s four driverless crashes are minor, and some weren’t Tesla’s fault. No one is denying that. But minor crashes still count, and so do the four Level 2 crashes that Tesla’s “zero injuries” framing leaves out. Both datasets describe the same vehicle running Tesla’s self-driving software.

It's looked to me for a while now that many of the improvements that people feel when they use FSD is not just the system getting better, but Elon pulling off the guardrails so the system attempts to do more stuff.

Recent findings suggest that what the data reveals is that Tesla has a speed problem. Its Robotaxi service looks clean largely because Tesla keeps it slow, off highways, and inside small zones. That’s also what makes it far less useful than a real ride-hailing service. You can’t take a Cybercab to the airport if the airport is on the other side of a freeway.

To be clear, we think Tesla is right to do it this way. Given what the Level 2 data highlights at 50+ mph, letting driverless Cybercabs onto highways today would be reckless.

But it means the hard part is still ahead. Highways, larger geofences, and higher speeds are where crashes become fatal, and that’s where Tesla needs its system to be nearly perfect before it can scale. The Cybercab’s test crashes suggest it isn’t there yet. The question is how long Tesla can keep growing a robotaxi network that can’t go fast.

Recent findings suggest that if you’re thinking about an EV, pairing it with home solar can cut your fuel costs to nearly zero. With electricity rates up almost 10% last year and expected to keep climbing, going solar is one of the best ways to protect yourself against rising costs. And with lease and PPA options, you can do it with zero upfront cost and start saving immediately. If you want to find the best deal, check out EnergySage. It’s a free service with hundreds of pre-vetted installers competing for your business, so you save 20 to 30% compared to going it alone. No sales calls until you pick an installer. Get your free quotes here.

FTC: We use income earning auto affiliate links. More.

According to recent reports, tesla is a transportation and energy company. It….

Fred is the Editor in Chief and Main Writer at Electrek.

You can send tips on Twitter (DMs open) or via email: fred@9to5mac.com.

Through Zalkon.com, you can check out Fred’s portfolio and get monthly green stock investment ideas.

Get an edge on MMA with the best stats.

EnergySage helps you get the best price possible on a home solar installation for free and without hassel.

→ Read original article