Musk Confirms Tesla FSD V14.3 Is in Testing, Wide Release Soon

Tesla owners have been eagerly awaiting the next major FSD release, and it looks like that wait may be almost over. Last night, Elon Musk officially confirmed that FSD v14.3 is being tested internally.

He went on to say that he expects it to go into wide release in a “few weeks.”

The Long Road from December

If the FSD v14.3 nomenclature sounds familiar, it’s because this specific build has been on the radar for quite some time. Originally, Elon had mapped out a software roadmap that slated v14.3 for a December 2025 launch.

However, as is often the case with the complexities of solving real-world AI, Tesla pivoted. Instead of dropping a large change with v14.3 at the end of last year, the Tesla fleet received a long series of v14.2.x bug fix builds. 

These iterative updates brought minor refinements to the vehicle’s driving behavior, prioritizing stability over sweeping architectural changes. Now, it seems the Tesla AI team is finally ready to push the true v14.3 overhaul to the public.

What Will FSD 14.3 Bring?

Musk hasn’t yet mentioned what would be included in the release. At the top of the wishlist is Banish. This is the next evolution of Autopark, which would allow your vehicle to drop you off at the door of your destination and autonomously navigate a crowded parking lot to find a spot and park itself.

The second item on the wishlist is improvements to Actually Smart Summon and Basic Summon, which are both built on the older FSD v12 stack and Autopilot stacks, respectively. Since these features operate on much older logic, they’re not as effective, smooth, or intelligent as FSD is in general.

Transitioning them to the latest end-to-end neural network architecture would be a massive upgrade for two of the most defining autonomy features and dramatically improve their reliability, speed, and safety.

Back in December, Musk famously said that FSD 14.3 would be the release that would make your car feel “sentient.” Back when Tesla released FSD 14.1, Musk stated that many more improvements were still to come in FSD v14, possibly hinting at new functionality. V14.3 could very well be the version that adds capabilities like Banish, rather than just refining existing features.

At this point in FSD’s cycle, most of the features needed to drive a car autonomously have been implemented. Tesla slowly pieced it together with highway driving, city streets, shifting between drive and reverse, and more recently, auto parking at the destination.

Tesla now needs to improve them to the point where the vehicle can operate safely with no one in it, which would then open the door to including banish.

Improved Reasoning Capabilities

Beyond parking lot tricks, v14.3 is expected to introduce a major change in how the vehicle actually thinks while navigating city streets.

Tesla’s Vice President of AI Software, Ashok Elluswamy, recently confirmed that reasoning was already implemented for FSD v14.2. However, v14.3 is supposed to bring another leap in the neural net’s ability to handle complex, unseen edge cases by actively reasoning through the scenario rather than just reacting to immediate obstacles.

As Tesla aggressively pushes toward its stated 10-billion-mile goal for unsupervised FSD, v14.3 will likely feature these improved reasoning capabilities more prominently. By allowing the car to make more human-like, contextual decisions, this update could be a bridge between the current supervised system and the autonomous Robotaxi future.

Tesla is wasting no time moving from promise to production. Just days after Elon Musk teased the official launch of the “Terafab” project, the company has already started the hunt for the specialized talent needed to build a semiconductor factory from the ground up.

The move was first spotted by X user @TeslaRyanRogue, who highlighted a new job listing for a Technical Program Manager in Infrastructure Semiconductors based in Austin, Texas. This signals that Tesla is moving past the conceptual stage and into the brutal reality of high-tech construction.

Why Tesla Wants to Make Its Own Chips

For a decade, Tesla has been a trailblazer in vertical integration, but computer chips remained one of the few critical components it had to outsource. As the company pivots from a pure carmaker to an AI and robotics giant, that dependency has become a massive liability. Musk has identified chip supply as the next major bottleneck for the company, especially as it prepares to mass-produce the Cybercab robotaxi and the Optimus humanoid robot.

The official job listing makes the scale of this project clear. Tesla is looking for someone to “own end-to-end program scoping — including factory design/construction from concept through execution, ramp-up, and production readiness.” Building a chip fab is arguably the hardest manufacturing challenge on Earth. Geographic location is vital because these facilities depend on incredibly precise lithography; even tiny vibrations from a nearby highway can ruin a production run. By moving this in-house in Austin, Tesla is attempting to do for silicon what it did for battery cells.

The Roadmap of Tesla Silicon

Tesla’s AI chips are the computational backbone for its entire ecosystem. Currently, the company relies on Samsung to manufacture its AI4 chips, which power the Full Self-Driving (FSD) system in mainstream vehicles today. The upcoming AI5 chip is already designed and will be built jointly by TSMC and Samsung, promising a generational leap in performance when it hits mass production in mid-2027.

Looking further ahead, Tesla has already signed a huge deal with Samsung to produce AI6 in U.S.-based fabs. However, the Terafab is the “endgame.” It would allow Tesla to iterate on future chip designs at its own pace. Musk has even suggested that the AI7 chip and beyond could eventually be deployed in orbital data centers in space, working alongside SpaceX and xAI.

Ramping Up the “Machine that Builds the Machine”

Building a fab isn’t just about putting up walls; it’s about gathering the equipment and expertise needed for complex chip fabrication processes like lithography, etching, and deposition in one place. Tesla is looking for leaders who have managed over $100 million in capital expenditures to oversee everything from utility planning to “tool installation” and “production qualification.”

This could very well be Tesla’s most audacious plan yet. While the company will continue to rely on TSMC and Samsung for the next few years, the Terafab represents a future where Tesla controls every single electron in its AI stack. If they can pull off a successful chip production ramp in Austin, the “bottleneck” of global chip supply will finally be a thing of the past for Musk’s empire.

Tesla is entering the home stretch as it prepares to launch the Cybercab, and the latest production units are showing off some major refinements. New sightings of the autonomous two-seater have revealed a significantly updated charge port and a thoughtful focus on accessibility that could make it a game-changer for urban mobility.

As Tesla prepares for volume production at Gigafactory Texas next month, these tweaks suggest the vehicle is moving away from some of its more experimental concepts in favor of real-world reliability.

A New and Improved Charge Port

The biggest hardware surprise comes from the charging system. While Tesla has long teased that the Cybercab would rely entirely on wireless induction charging, recent production units have been spotted with a much more “finished” physical charge port.

According to JoshWest247 on X, the new port features several “huge changes” compared to earlier prototypes. The design is now fully integrated with a simpler door, a full surround gasket, and a flexible housing that eliminates the need for a weather plug. It’s significantly different and a lot more polished than the previous iteration, which we saw in action last month.

Importantly, the flap on the new charge port isn’t motorized, likely to optimize for low manufacturing and replacement costs. Seeing such a fleshed-out physical port this close to the April production start suggests that the Cybercab will launch with traditional plug-in charging first, using a native NACS port. While Tesla recently won FCC approval for the Ultra Wideband technology needed for wireless charging, that feature now looks like an “eventually” upgrade rather than a day-one necessity.

Built for Accessibility

Beyond charging, Tesla is also focusing on who will be using these vehicles. Eric E., Tesla’s Cybercab Engineering Lead, recently confirmed that the vehicle was designed with accessibility at its core.

“We built the seat height to be inline with standard wheel chair seat height. Accessibility is very important in autonomous vehicles, freedom of transportation for everyone is critical,” Eric said on X.

By matching the seat height to a standard wheelchair, Tesla is making it much easier for passengers with limited mobility to slide into the cabin, a crucial feature for a vehicle meant to provide “freedom of transportation for everyone.” According to the Tesla engineer, the Cybercab’s seat height and unique butterfly doors allow for a wheelchair to be parked parallel to the seat with plenty of room to maneuver between the two.

Production Refinements and Final Specs

The first production Cybercab has already rolled off the line, and we are seeing a massive “sanity check” of the hardware. Recent sightings in Austin have revealed larger front-facing cameras, an interior trunk camera to ensure passengers don’t leave anything behind, and even ambient lighting all along the cabin.

To ensure the Full Self-Driving system stays active in all conditions, the Cybercab is equipped with a high-pressure washer system for every external camera. This is a level of redundancy not seen on consumer Teslas, and it’s necessary for a vehicle the company plans to eventually ship without a steering wheel or pedals. Inside, nearly every function is handled by a massive 21-inch touchscreen, with only physical buttons for the doors and windows, along with a hazards button that doubles as an emergency stop trigger.

As Tesla moves toward its April mass production target, the Cybercab is looking less like a science project and more like a rugged, accessible workhorse for the upcoming Robotaxi network. The transition to a driverless society is quickly becoming a reality, and it seems Tesla is making sure everyone — including those with mobility challenges — can come along for the ride.