If you've searched for what Teslas have Full Self-Driving, you've likely hit a wall of conflicting claims. Tesla markets "Full Self-Driving (Supervised)" as an optional software package, but the name is aspirational — not a description of what the system actually does today. Every current-production Tesla ships with the hardware necessary to run FSD Supervised, but none of them drive themselves. Here's the precise breakdown of which models carry the latest hardware, what FSD Supervised can and cannot do, and how Tesla's approach compares to genuine Level 4 robotaxi services like Waymo.
Every Current Tesla Has FSD Hardware — But That's Not the Whole Story
As of the 2026 model year, every new Tesla — Model 3, Model Y, Model S, Model X, and the Cybertruck — ships from the factory with Tesla's latest AI hardware suite (commonly referred to as Hardware 4 or HW4). This includes:
- 8 cameras (front, side repeaters, rear) providing 360-degree vision
- No lidar, no radar (Tesla removed ultrasonic sensors from new production starting late 2022, relying on Tesla Vision camera-only perception)
- An onboard computer running Tesla's custom FSD neural network inference chip
However, having the hardware is only half the equation. FSD (Supervised) is a paid software upgrade — currently priced at $8,000 as a one-time purchase or available via a $99/month subscription in the U.S. Without purchasing FSD, your Tesla ships with Basic Autopilot (adaptive cruise + lane centering on highways) or Enhanced Autopilot ($6,000, adding auto lane change, navigate on Autopilot, auto park, and Summon).
Tesla's FSD (Supervised) and Autopilot are SAE Level 2 driver-assistance systems. This means the vehicle can control steering and acceleration simultaneously, but the driver must remain fully attentive, hands on or near the wheel, and ready to intervene instantly. This is fundamentally different from SAE Level 4 robotaxi services (Waymo, Zoox), which operate without a human driver in defined areas. No Tesla sold today is Level 3, Level 4, or Level 5 autonomous.
Which Tesla Models Support FSD Supervised? Full 2026 Lineup
| Model | Hardware | FSD Eligible | Starting MSRP | EPA Range |
|---|---|---|---|---|
| Model 3 (Long Range AWD) | HW4 (AI4) | Yes | ~$47,740 | 341 mi |
| Model 3 (Performance) | HW4 (AI4) | Yes | ~$52,990 | 315 mi |
| Model Y (Long Range AWD) | HW4 (AI4) | Yes | ~$48,990 | 320 mi |
| Model Y (Performance) | HW4 (AI4) | Yes | ~$53,990 | 285 mi |
| Model S (Long Range) | HW4 (AI4) | Yes | ~$79,990 | 402 mi |
| Model X (Long Range) | HW4 (AI4) | Yes | ~$84,990 | 335 mi |
| Cybertruck (AWD) | HW4 (AI4) | Yes | ~$79,990 | 340 mi |
Important caveat for older Teslas: Vehicles built before April 2019 carry Hardware 2.0 or 2.5, which Tesla has confirmed cannot run the current FSD neural network. Owners of those vehicles would need a hardware retrofit (previously offered at no cost to FSD purchasers, but availability varies). Vehicles with Hardware 3.0 (built roughly 2019–2023) can run FSD Supervised but may not support future compute-intensive updates without an upgrade to HW4.
What FSD Supervised Actually Does in 2026
FSD (Supervised) extends Tesla's Autopilot beyond highways to city streets. The current feature set includes:
- Traffic light and stop sign recognition — the car slows and stops at controlled intersections, proceeding on green after driver confirmation
- City street autosteer — the vehicle follows a navigation route on surface streets, making turns, navigating roundabouts, and handling lane changes
- Auto lane change on highways — initiated by the driver via turn signal or suggested by the system
- Smart Summon — the car navigates a parking lot to the driver's phone location at low speed
- Autopark — parallel and perpendicular parking with minimal input
The system relies entirely on Tesla's end-to-end neural network architecture (introduced in FSD v12 and refined through subsequent versions). Rather than hard-coded rules for every scenario, the neural network was trained on billions of miles of real-world driving video captured from the Tesla fleet. This approach produces more human-like driving behavior but also means edge cases — unusual construction zones, erratic pedestrians, severe weather — can produce unpredictable responses.
- Drive without a licensed, attentive driver in the seat
- Operate safely in heavy rain, snow, fog, or low-visibility conditions (camera-only system degrades significantly)
- Handle all construction zones, unmapped road changes, or complex multi-point turns reliably
- Replace driver judgment — you are legally responsible for every action the car takes
- Operate as a robotaxi — Tesla's planned "Robotaxi" service remains announced but not commercially available to the public as of 2026
Tesla FSD vs. True Robotaxis: The SAE Level Divide
Understanding what Teslas have Full Self-Driving requires understanding that Tesla's system and services like Waymo occupy entirely different categories of autonomy. Here's how they compare:
| Attribute | Tesla FSD (Supervised) | Waymo (Level 4) | Zoox (Level 4) |
|---|---|---|---|
| SAE Level | Level 2 (driver-assist) | Level 4 (driverless in geofence) | Level 4 (driverless in geofence) |
| Human Driver Required | Yes — always | No | No |
| Sensor Suite | Camera-only (8 cameras) | Lidar + radar + cameras | Lidar + radar + cameras |
| Operational Area | Most public roads (supervised) | Geofenced metro areas | Limited geofenced zones |
| Weather Capability | Degraded in rain/snow (no lidar) | Better in adverse weather (lidar redundancy) | Better in adverse weather (lidar redundancy) |
| Cost to Consumer | $8,000 purchase or $99/mo + vehicle cost | Per-ride fare (comparable to Uber/Lyft) | Per-ride fare (limited availability) |
| Liability | Driver is liable | Waymo assumes liability in-operation | Zoox assumes liability in-operation |
Where Robotaxi Services Actually Operate Today
If you're researching Tesla FSD because you want a driverless ride, you need a robotaxi — not a Level 2 system. Here's where genuine driverless services are live as of 2026:
| Service | Live Cities (Driverless) | Hours | Status |
|---|---|---|---|
| Waymo One | Phoenix, San Francisco, Los Angeles, Austin | 24/7 in most markets | Commercially open to public |
| Zoox | Las Vegas (limited), Foster City, CA (testing) | Limited hours | Invite-only / early access |
| Cruise | Operations suspended/restructuring post-2023 incident | N/A | Phased relaunch with safety driver |
| Tesla Robotaxi | None (announced, not commercially launched) | N/A | Concept / future service |
Safety Data: What the Numbers Actually Show
Any discussion of what Teslas have Full Self-Driving capability must address safety honestly. Tesla publishes periodic safety reports, but methodology matters:
Tesla's quarterly safety reports show approximately one crash per ~4–5 million miles driven with Autopilot engaged, compared to the U.S. average of roughly one crash per ~670,000 miles (NHTSA national average). However, these numbers require context:
- Autopilot is primarily used on highways, which are statistically safer than city streets
- Tesla's fleet skews newer with advanced passive safety features
- FSD Supervised on city streets has a higher intervention rate than highway Autopilot
- Comparisons to the national average don't control for road type, driver age, or vehicle age
For robotaxi comparison, Waymo has published data through its safety research showing over 30 million fully driverless miles completed with a lower rate of police-reported crashes compared to human benchmarks in the same geographies. Waymo's redundant sensor suite (lidar, radar, cameras) provides detection capability that Tesla's camera-only system cannot match in low-visibility scenarios.
The Technology Stack: Camera-Only vs. Multi-Sensor
Tesla's bet on a camera-only (Tesla Vision) approach is the most controversial decision in autonomous driving. Here's what each side offers:
Tesla Vision (Camera-Only)
- Pros: Lower hardware cost, scalable to millions of consumer vehicles, rich color/texture data for neural networks, mimics human vision
- Cons: No direct distance measurement (relies on learned depth estimation), degrades in rain/fog/glare/darkness, single-point-of-failure risk
Lidar + Camera + Radar (Waymo, Zoox, Cruise)
- Pros: Lidar provides precise 3D distance regardless of lighting, radar penetrates weather, redundancy across sensor types, better performance at night and in adverse conditions
- Cons: Higher cost ($10,000–$100,000+ per vehicle sensor suite), not yet practical for consumer vehicles at scale, larger physical package
The practical implication: Tesla FSD Supervised performs well in clear daylight on well-marked roads but requires immediate driver takeover in scenarios where a lidar-equipped robotaxi would continue operating autonomously. This is why Tesla mandates supervision — the system's sensor limitations create genuine blind spots that only a human driver can cover.
Regulatory Landscape: Who Approves What
In the United States, autonomous vehicle regulation is split between federal and state authority:
- NHTSA (federal): Sets vehicle safety standards and investigates defects. NHTSA has opened multiple investigations into Tesla Autopilot/FSD crashes, particularly involving stationary emergency vehicles. NHTSA does not "approve" autonomous features — it regulates after-the-fact through recalls and enforcement.
- State regulators: Control where and how driverless vehicles can operate commercially. California (DMV and CPUC), Arizona, Texas, and Nevada have issued the most robotaxi permits. Tesla's FSD Supervised does not require special state permits because it's classified as driver-assist, not driverless.
- FTC / consumer protection: Tesla has faced scrutiny from consumer advocates and lawmakers over the "Full Self-Driving" name, which critics argue is misleading under consumer protection statutes.
Tesla's planned robotaxi service — where consumers hail a driverless Tesla — would require the same state-level commercial permits that Waymo holds. As of 2026, Tesla does not hold commercial driverless permits in any U.S. jurisdiction.
Is FSD Worth $8,000? A Practical Decision Framework
Buy FSD Supervised If:
- You commute long distances on highways and want the most advanced lane-centering + auto lane-change system available on a Tesla
- You enjoy being an early adopter of neural-network driving tech and understand the system will make mistakes requiring your intervention
- You plan to keep the vehicle long enough that the $99/month subscription total would exceed $8,000 (about 6.7 years)
- You believe Tesla may eventually achieve higher autonomy levels and want hardware compatibility
Skip FSD If:
- You expect a self-driving car — FSD Supervised is not one, and no timeline for true autonomy has been met
- You primarily drive in heavy rain, snow, or poorly-marked rural roads where camera-only systems struggle most
- You want a system where the manufacturer assumes liability — with FSD, you are always the responsible driver
- Basic Autopilot (included) or Enhanced Autopilot ($6,000) covers your use case
Frequently Asked Questions
Can any Tesla drive itself without a driver?
No. Every Tesla sold, including all 2026 models with HW4, requires an attentive, licensed driver in the driver's seat at all times when FSD (Supervised) or Autopilot is engaged. Tesla's announced "Robotaxi" concept vehicle has not launched as a commercial service and has no confirmed public availability date.
What's the difference between Autopilot, Enhanced Autopilot, and FSD?
Basic Autopilot (included with every Tesla) provides adaptive cruise control and lane-centering on highways. Enhanced Autopilot ($6,000) adds auto lane change, Navigate on Autopilot (highway on/off-ramp guidance), auto park, and Smart Summon. FSD (Supervised) ($8,000 or $99/month) adds city-street autosteer with traffic light/stop sign recognition and surface-street navigation. All three are SAE Level 2 — the driver must supervise.
Will older Teslas get FSD updates?
Teslas with Hardware 3.0 (roughly 2019–2023 production) can currently run FSD Supervised. Tesla has indicated that future FSD versions may require more compute than HW3 provides, potentially requiring a paid hardware upgrade. Vehicles with HW2.0 or HW2.5 (pre-April 2019) cannot run the current FSD neural network at all.
How does Tesla FSD compare to Waymo for getting a ride without driving?
They're not comparable for that purpose. If you want a ride without driving, you need a Level 4 robotaxi like Waymo One (available in Phoenix, San Francisco, Los Angeles, and Austin). Tesla FSD always requires you behind the wheel. Waymo fares are generally comparable to Uber/Lyft — typically $10–$25 for a typical urban trip depending on distance and demand.
Does FSD improve my Tesla's resale value?
Historically, FSD has not transferred significant value to resale. Third-party data from sources like iSeeCars shows FSD-equipped used Teslas sell for only a modest premium over non-FSD equivalents — far less than the $8,000 purchase price. Treat FSD as a personal-use expense, not an investment.
Is Tesla's FSD safe in rain or snow?
Camera-only systems degrade in precipitation, fog, and low-light conditions. Tesla's owner manual explicitly warns drivers to exercise increased caution and be prepared to take over immediately in adverse weather. Lidar-equipped robotaxis handle these conditions better due to sensor redundancy, though even they may limit operations in severe weather.
