HW1 / AP1
2014-2016- Computer
- Mobileye EyeQ3
- Sensors
- 1 camera + radar + ultrasonics
- Key change
- First Tesla Autopilot
Tesla hardware guide
A complete buyer-focused guide to Tesla Autopilot and Full Self-Driving hardware, from HW1 and Mobileye to HW4 and future AI5 systems.
Tesla vehicles can look identical externally while carrying very different Autopilot and FSD hardware. That difference can change what a buyer should expect from the car.
Tesla does not always change hardware exactly on model-year boundaries, so use these ranges as practical buyer context rather than a substitute for checking the actual car.
| Hardware | Approx. years | Computer | Sensors | Key change |
|---|---|---|---|---|
| HW1 / AP1 | 2014-2016 | Mobileye EyeQ3 | 1 camera + radar + ultrasonics | First Tesla Autopilot |
| HW2 / AP2 | 2016-2017 | NVIDIA Drive PX2 | 8 cameras + radar + ultrasonics | Tesla-designed perception stack |
| HW2.5 | 2017-2019 | Enhanced NVIDIA platform | 8 cameras + radar + ultrasonics | More redundancy, dashcam, Sentry Mode support |
| HW3 / FSD Computer | 2019-2023 | Tesla custom FSD computer | Cameras + radar + ultrasonics initially | Major AI compute leap |
| Tesla Vision HW3 | 2021-2023 | Tesla FSD computer | Cameras only on many cars | Radar removal / camera-first autonomy |
| Vision without USS | 2022-present | Tesla FSD computer / AI4 | Cameras only | Ultrasonic parking sensor removal |
| HW4 / AI4 | 2023-present | Tesla next-gen FSD computer | Higher-resolution cameras, updated sensor architecture | Major compute and camera upgrade |
| HW5 / AI5 | Future / announced | Next-gen Tesla AI computer | TBD | Future autonomy platform |
Use these sections to translate seller language into the hardware era that matters for a used Tesla decision.
HW1 cars introduced many buyers to Tesla Autopilot. They can still be enjoyable used Teslas, but their hardware belongs to an earlier driver-assistance era.
Buyer takeaway
HW1 cars can be great used Teslas, but they are legacy Autopilot vehicles and should not be treated as modern FSD-capable cars.
Late-2016 AP2 cars marked a major architectural reset as Tesla moved away from Mobileye and began building around its own neural-network perception stack.
Buyer takeaway
HW2 was a major reset. Some cars may be upgradeable to HW3 if FSD was purchased, but verify the actual installed computer.
HW2.5 refined the AP2 platform before Tesla's own FSD computer arrived. It is common in early Model 3s and remains important in used Tesla evaluations.
Buyer takeaway
HW2.5 vehicles are often solid used buys, but FSD capability depends on whether the vehicle has been upgraded to HW3.
HW3 brought Tesla-designed FSD compute into production and became the core hardware era for many used Teslas with FSD-era software expectations.
Buyer takeaway
HW3 is the core FSD-era hardware for many used Teslas, but it is no longer the newest Tesla autonomy hardware.
Tesla Vision describes Tesla's move toward camera-first Autopilot and FSD perception. It matters because two similar-looking HW3 cars may differ in radar hardware or active radar use.
Buyer takeaway
Vision-only cars are common. Buyers should know whether a car has radar, has radar disabled, or was built without radar.
The removal of ultrasonic sensors changed the parking-assist story for many newer Teslas. This is one of the easiest hardware differences to miss in listing photos.
Buyer takeaway
For buyers who care about precise parking assist, Summon, or Smart Summon behavior, USS status matters.
HW4 is the newer production autonomy hardware generation and is usually the strongest future-facing signal for buyers comparing late-model Teslas.
Buyer takeaway
HW4 is currently the most desirable Tesla autonomy hardware generation for future-facing buyers.
Tesla has discussed AI5/HW5 as a future-generation inference computer, but it should not be treated as a normal used-market feature today.
Buyer takeaway
Do not assume a used Tesla has HW5 unless there is explicit evidence. Treat HW5 as future-looking, not current used-market standard.
The biggest buyer-facing changes are easier to understand when cameras, radar, and ultrasonic sensors are separated.
Keep these guides nearby while you compare hardware generation, battery context, and VIN-specific claims.
Tesla research
Open Tesla make, model, and year VIN decoder pages for trim, battery, build-date, and option checks.
Explainer
Learn Tesla-specific VIN digit meanings, model-year, factory, trim, battery, and Autopilot hardware checks.
Technical guide
Add battery degradation and range-context checks to your vehicle research without overreacting to one metric.
Technical guide
Interpret Tesla hardware and option-package claims with cleaner model-year context.
Next step
Speckr decodes Tesla VINs using deep vehicle data so buyers and dealers can understand the real configuration behind the listing.
Cookie preferences
We use first and third-party cookies and similar technologies to collect website usage data and improve, protect, and promote our products and services. You can choose to accept all cookies or accept only necessary cookies.