AUSTIN, Texas — Tesla has begun rolling out Full Self-Driving (Supervised) v14.1 Lite, a release that brings the intelligence of its latest HW4 software down to millions of older Hardware 3 vehicles and adds a wave of new capabilities for owners.
Shipping as part of software update 2026.21.6 with a release date of Aug. 9, the update is significant because it narrows the gap between Tesla's two generations of self-driving hardware. As the official release notes explain, Tesla "distilled the intelligence from HW4 V14 into HW3," letting older cars directly learn how to handle scenarios using the newer stack as a guide.
Bringing V14 to Older Cars
For years, HW3 owners have watched newer vehicles receive the flashiest FSD improvements. V14.1 Lite changes that by unlocking gains that had been exclusive to HW4, including advances from reinforcement learning and offline models. Tesla says the update improves both proactive and reactive responsiveness across navigation, merges and forks, pedestrian interactions, traffic lights, and vehicle cut-in scenarios, while also smoothing steering and lane centering for greater everyday comfort.
The emphasis on learning-based driving reflects the AI-first approach Tesla has embraced across the company, a strategy captured in our coverage of how Musk put AI and robots at the center of Tesla's story.
New Features for Drivers
Beyond raw capability, v14.1 Lite introduces practical tools. HW3 cars now gain parking, unparking, and reversing abilities, along with new Arrival Options that let drivers choose where FSD should pull over or park — in a parking lot, on the street, or in a driveway. A new Speed Profiles feature lets owners tune driving style, including a more conservative "SLOTH" profile that drives even more cautiously than the existing "CHILL" setting.
These refinements build on Tesla's broader effort to make FSD feel more personal, an approach seen recently when Tesla tuned FSD to learn each driver's style. The companion HW4 release, v14.3.7, layers on further gains, including a rewritten AI compiler that Tesla says delivers 20% faster reaction time.
A Steady March Toward Autonomy
The rollout is still in its early phase, seeding to a small slice of the fleet before wider distribution, as Tesla monitors real-world performance and irons out any issues. But the direction is clear: by distilling its best models down to hardware already on the road, Tesla is extending the value of vehicles owners bought years ago and keeping its enormous installed base on a single, improving software trajectory.
That approach — squeezing more capability out of existing hardware through software and AI — is central to Tesla's bet on autonomy. Every incremental update pushes the fleet closer to the company's long-stated goal of supervised self-driving that works almost everywhere, and v14.1 Lite ensures HW3 owners are along for the ride.
