News
Tesla FSD Beta V11.4.1 has ‘major architectural improvements’: Elon Musk
Tesla’s release notes for FSD Beta V.11.4.1 are out now.
Version 11.4.1 may seem like another minor update, but Elon Musk commented that it introduces some significant improvements to FSD Beta. In a tweet, Musk hinted about what’s to come in FSD Beta V.12.
“As mentioned earlier, V11.4.1 has major architectural improvements it’s actually much more than a point release. Should arguably be v12.0, but that reserved for when FSD is fully AI from video in to control out,” Musk commented.
Below are all the improvements introduced in V11.4.1.
- Improved control through turns, and smoothness in general, by improving geometry, curvature, position, type and topology of lanes, lines, road edges, and restricted space. Among other improvements, the perception of lanes in city streets improved by 36%, forks improved by 44%, merges improved by 27% and turns improved by 16%, due to a bigger and cleaner training set and updated lane-guidance module.
- Added lane-guidance inputs to the Occupancy Network to improve detections of long-range roadway features, resulting in a 16% reduction in false negative median detections.
- Improved ego’s assertiveness for crossing pedestrians in cases where ego can easily and safely cross before the pedestrian.
- Improved motorbike recall by 8% and increased vehicle detection precision to reduce false positive detections. These models also add more robustness to variance in vision frame-rate.
- Reduced interventions cause by other vehicles cutting into ego’s lane by 43%. This was accomplished by creating a framework to probabilistically anticipate objects that may cut into ego’s lane and proactively offset and/or adjust speed to position ego optimally for these futures.
- Improved cut-in control by reducing lane-centric velocity error by 40-50% for close-by vehicles.
- Improved recall for object partial lane encroachment by 20%, high yaw-rate cut-in by 40%, and cut-out by 26% by using additional features of the lane-change trajectory to improve supervision.
- Reduced highway false slowdowns related to underestimated velocities for faraway object by adding 68K videos to the training set with improved auto-labeled ground truth.
- Smoothed in-lane offsetting for large vehicles by tuning the amount of lateral jerk allowed for the maneuver.
- Improved lateral control for upcoming high-curvature merges to bias away from the merging lane.
Do you notice a change in V11.4.1? I’d like to hear your thoughts on the FSD beta update. Please get in touch with me at maria@teslarati.com or via Twitter @Writer_01001101.
Elon Musk
SpaceX wants to catch Starship for launch 14, Elon Musk says
Just hours after Starship Flight 13 achieved a successful soft splashdown of its upper stage in the Indian Ocean on July 24, Elon Musk announced an ambitious next step for the company’s next launch of the rocket.
“Unless we discover problems after mission data review, SpaceX will attempt to catch the ship with the tower on [the] next flight,” the SpaceX CEO posted on X on Friday.
That “next flight” is expected to be Flight 14. The plan involves returning the Starship upper stage, commonly called the “ship,” to the Starbase launch tower in Texas and catching it mid-air using the same mechanical “chopsticks” arms that have already proven themselves with the Super Heavy booster.
Unless we discover problems after mission data review, SpaceX will attempt to catch the ship with the tower on next flight
— Elon Musk (@elonmusk) July 25, 2026
A successful catch would mark the first time an orbital-class upper stage has been recovered this way, advancing SpaceX’s goal of full and rapid reusability for the entire vehicle.
SpaceX has already demonstrated the tower-catch technique multiple times with Super Heavy. The first successful catch came on Flight 5 in October 2024, when Booster 12 was plucked from the sky by the Mechazilla arms. Subsequent flights, including those involving Boosters 14 and 15, repeated the feat. Several of those recovered boosters were later inspected, refurbished, and flown again, proving the system’s viability for quick turnaround.
Traditional reusable rockets, such as SpaceX’s own Falcon 9 or Blue Origin’s New Shepard, land on legs either on land or droneships. Rocket Lab has recovered its small Electron first stages by helicopter, but those are far lighter vehicles.
SpaceX Starship just nailed something it’s never done before
The China Academy of Launch Vehicle Technology (CALT), a subsidiary of the China Aerospace Science and Technology Corp. (CASC), completed a catch of its booster on July 10. They are the only entity besides SpaceX to attempt and complete the feat.
Flight 13 provided encouraging data. The ship executed a controlled reentry, flipped, and soft-landed intact in the ocean after deploying Starlink satellites, offering the first clear post-splashdown views of an undamaged heat shield. The Super Heavy booster, meanwhile, experienced a harder splashdown in the Gulf of Mexico.
Musk has previously stressed that ship catches would only follow multiple successful soft ocean landings to minimize risk of debris over land.
If Flight 14 succeeds, SpaceX would take a major stride toward routine, rapid reuse of both stages—critical for lowering launch costs and supporting ambitious plans for lunar and Mars missions. For now, teams are reviewing the Flight 13 data. Should everything check out, the next Starship flight could deliver one of the most spectacular recoveries in aerospace history.
News
Tesla to open source Model S and Model X designs and software
In a move echoing its earlier commitment to open innovation, Tesla CEO Elon Musk announced recently that the company plans to make the design and software of its Model S and Model X fully open source.
This follows the same approach Tesla took with its original Roadster, releasing all available design, engineering, and diagnostic materials in November 2023 so that “whatever we have, you now have.”
Just as Tesla made the original Roadster design & software open source, we plan to do the same with Model S & X
— Elon Musk (@elonmusk) July 24, 2026
The Model S, introduced in 2012, was Tesla’s first mass-produced vehicle and a groundbreaking luxury electric sedan. It offered impressive range, rapid acceleration, and over-the-air software updates that redefined expectations for electric cars.
The Model X, launched in 2015, built on that foundation as a high-performance electric SUV notable for its distinctive falcon-wing doors, spacious interior, and advanced safety features. Both models served as flagships that helped establish Tesla as a leader in the EV industry and popularized long-range battery-electric vehicles.
Production of the Model S and Model X was wound down earlier in 2026, with manufacturing ending in the second quarter. Tesla redirected the Fremont factory space previously used for these vehicles toward higher-priority projects, including Optimus humanoid robots and the Cybercab autonomous vehicle.
By the time of Musk’s open-source announcement, custom orders had closed and only remaining inventory was available.
Open-sourcing the designs and software offers several clear advantages. Owners of these aging but still capable vehicles gain better access to technical documentation, diagnostic tools, and software resources, making independent repairs and modifications easier and more affordable.
Independent repair shops and third-party specialists can support the large existing fleet without relying solely on Tesla’s service network. Enthusiasts and engineers can study real-world implementations of Tesla’s battery, powertrain, and software systems, potentially accelerating broader industry progress in electric mobility.
The step aligns with Tesla’s 2014 patent pledge and its overall mission to advance sustainable transport by sharing hard-won knowledge rather than locking it behind proprietary walls.
By releasing these materials now that the models have left production, Tesla ensures continued support for its early adopters while freeing internal resources for future technologies. The open-source release of the original Roadster already enabled simulations, community projects, and deeper technical understanding.
Extending that practice to the Model S and Model X should deliver similar benefits on a larger scale, helping keep these influential vehicles relevant and repairable for years to come
News
Tesla flexes incredible Robotaxi metric that skeptics will hate
Tesla flexed one incredible Robotaxi metric during the Q2 Earnings Call that skeptics have to hate to hear. The company’s platform has already driven more than 380,000 miles of unsupervised ride-hailing across several states with no notable incidents.
During the company’s Q2 Earnings Call on Wednesday, Vice President of AI, Ashok Elluswamy, said:
“First of all, I’d like to state that the Robotaxi program has been operating extremely well. Especially in terms of safety, the program has had an impeccable safety record. We have driven more than 380,000 miles of unsupervised Robotaxi, now across six cities in two different states. We have had zero notable incidents. Any reports have been of other actors impacting us when we were stationary. I like to emphasize how safe the operation has been so far. Zero notable incidents over 380,000 miles.”
Elluswamy’s claim over Robotaxi miles is a significant milestone for Tesla in the grand scheme, especially considering this is a sizeable number of miles without any incident.
0 notable incidents across over 380,000 miles traveled by Robotaxi
— Tesla (@Tesla) July 22, 2026
Tesla’s self-driving approach is much different than that of other companies. Tesla has maintained that vision is the only thing needed to have a solid and effective self-driving suite. Many self-driving companies utilize things like LiDAR, sensors, and other elements to improve performance, but Elluswamy sent a jab at those who believe it’s needed.
“Historically, the so-called experts have always claimed that you need LiDARs, radars, HD maps, and the entire kitchen sink to drive safely. Here we show that such is not true. You can have safe, comfortable, and affordable autonomy with just cameras. This record should be a huge validation of Tesla’s entire AI approach.”
The feat of accumulating this many miles without any driver behind the wheel is impressive. The thing is, Tesla is also doing this across several different locations, with varying traffic rules, pedestrian levels, weather patterns, and other important factors.
While Tesla is not ready to roll out an unsupervised platform completely, it is a slow but steady indication that the company is well on its way to figuring things out.
The company’s attitude toward expansion is slow, safe, and controlled, and despite this huge milestone, it will still be some time until we see Tesla truly unleash unsupervised rides more aggressively.
