Full Self-Driving (Supervised)

What the package is

Full Self-Driving (Supervised) is the name Tesla uses for its most advanced driver-assistance software. The word in parentheses carries most of the weight: Tesla states that the features currently enabled require active driver supervision and do not make the vehicle autonomous ^[1]. The car can take over long stretches of the driving task, but the person in the driver's seat is expected to keep watching the road and be ready to take back control.

That framing matters for anyone reading the name literally. "Full Self-Driving" describes the scope of the driving task the software is meant to handle, not the level of autonomy the driver is granted today.

How it got here

The software first reached customers as a limited release. In October 2020 Tesla released Full Self-Driving (Beta) to a small set of users ^[2]. A beta in this sense is a staged rollout: a small group drives unfinished features on public roads, and the software changes between releases.

Tesla now sells Full Self-Driving (Supervised) as a subscription at $99 per month ^[1]. Selling the software monthly rather than only as a one-off purchase turns it into a continuing service, priced so that owners can switch it on for a season and off again.

What the car actually does

Tesla lists the enabled features as route navigation, steering, lane changes and parking ^[1]. On a planned trip, that combination means the car can follow a route, hold a lane and move between lanes, with the driver monitoring throughout.

Two of the listed functions deal with parking, where low speeds make the task more contained. Actually Smart Summon lets the car navigate parking lots to reach its owner, who calls it from the app ^[1]. Autopark detects perpendicular and parallel parking spots and manoeuvres into them ^[1]. Both are narrow, well-defined jobs compared with driving through traffic, which is why they are described as separate features rather than as part of the general driving function.

Where it works

The system is available in select markets across North America, Europe and Asia Pacific, with expansion expected in future updates ^[1]. Availability therefore differs by country even within those regions; a feature that is switched on in one market is not automatically switched on in another.

Where the training comes from

Tesla says the system is trained on what amounts to over 100 years of anonymous real-world driving scenarios, collected from a fleet of more than six million vehicles ^[1]. The company also says its fleet collectively experiences a lifetime of driving scenarios every 10 minutes ^[1].

That scale is the core of Tesla's argument for its approach. Rather than writing explicit rules for every situation a car might meet, the system learns from what its own vehicles encounter on real roads, and the fleet supplies those encounters continuously. The word "anonymous" matters here: the claim is about driving situations, not about identifying the people in the cars that recorded them.

What comes next

Tesla positions FSD (Unsupervised) as the next step, which it says will unlock its robotaxi fleet and the Cybercab ^[1]. The contrast between the two names is a useful way to read the product: supervised and unsupervised describe who is responsible for the driving, not how many features the software has. Until unsupervised operation arrives, what customers can buy remains an assistance package, with the driver in charge.