Connect with us

News

Elon Musk shares Tesla FSD Beta 9.2’s behind-the-scenes technical release notes

Credit: @brandonee916/Twitter

Published

on

Tesla released FSD Beta 9.2 over the weekend, and while the update’s official Release Notes were identical to FSD Beta 9.2, there were a lot of improvements that were rolled out by the company under the hood. These improvements were so notable that Elon Musk stated on Twitter that 9.2’s release had to be moved back a bit to address some last-minute issues. 

The Tesla CEO then decided to share the technical release notes for FSD Beta 9.2, which explains every optimization that has been introduced in the recent update. Based on what Musk has shared, it appears that FSD Beta 9.2 is a massive step up from V9.1, allowing the advanced driver-assist system to navigate through inner-city streets in a safer and more confident manner. 

The following are the technical release notes for FSD Beta 9.2, as shared by CEO Elon Musk on Twitter. 

  • Clear-to-go boost through turns on minor-to-major roads (plan to expand to all roads in V9.3).
  • Improved peek behavior where we are smarter about when to go around the lead vehicle by reasoning about the causes for lead vehicle being slow.
  • V1 of the Multi-modal prediction for where other vehicles expected to drive. This is only partially consumed for now.
  • New Lanes network with 50k more clips (almost double) from the new auto-labeling pipeline.
  • New VRU velocity model with 12% improvement to velocity and better VRU clear-to-go performance. This is the first model trained with “Quantization-Aware-Training”, an improved technique to mitigate int8 quantization.
  • Enabled Inter-SoC synchronous compute scheduling between vision and vector space processes. Planner in the loop is happening in v10.
  • Shadow mode for new crossing/merging targets network which will help improve VRU control. 

Apart from the technical details of FSD Beta 9.2’s release, Elon Musk also hinted at the upcoming updates coming for the advanced driver-assist system. The CEO noted that FSD Beta 9.2 would be followed by V9.3 and V9.4, but after this point, Tesla may already release FSD Beta 10. V10 would feature “significant architecture changes,” as per the CEO. Musk did not explain further, though speculations suggest that FSD Beta 10 would be the update that would transition Tesla into a single FSD stack

Lastly, Musk provided some updates on the release date of the FSD Beta Download Button, which has experienced substantial delays. The CEO explained that the “Button” would probably roll out with Beta 10 or V10.1, which, based on the current release cycles of FSD Beta’s updates, would likely be in the next couple of months, barring any unexpected delays.  

Advertisement

“Beta 10 or maybe 10.1. Going to pure vision set us back initially. Vision plus (coarse) radar had us trapped in a local maximum, like a level cap. Pure vision requires fairly advanced real-world AI, but that’s how our whole road system is designed to work: NN’s with vision,” Musk explained. 

Don’t hesitate to contact us with news tips. Just send a message to tips@teslarati.com to give us a heads up. 

Simon is an experienced automotive reporter with a passion for electric cars and clean energy. Fascinated by the world envisioned by Elon Musk, he hopes to make it to Mars (at least as a tourist) someday. For stories or tips--or even to just say a simple hello--send a message to his email, simon@teslarati.com or his handle on X, @ResidentSponge.

Advertisement
Comments

News

Tesla and driver sued by family of woman killed in Texas crash: what we know

Published

on

Credit: CNBC

Tesla is being sued by the family of the woman who was killed in a Texas crash involving a Model 3. The driver, who is also being sued, claimed the vehicle was operating on Autopilot mode, but Tesla executives have come out challenging that claim, stating that the driver of the vehicle overrode the system.

The lawsuit was filed by 76-year-old Martha Avila’s daughter and her husband, who allege a “design defect” involving a Tesla and a failure to warn. The suit alleges negligence against Tesla and the driver, Michael Butler.

Butler “stated he was operating with an automated driving assistance system engaged at the time of the crash,” the Harris County Sheriff’s Office said in a statement. He showed no signs of intoxication and was cooperative, the Sheriff’s Office said, according to NBC News.

Just after reports of the crash and numerous headlines that immediately blamed Tesla’s Autopilot suite, both Tesla CEO Elon Musk and Head of AI Ashok Elluswamy challenged that. Musk said the crash made “no sense” given that Tesla Autopilot and Full Self-Driving do not travel at the speeds the door cameras captured the car traveling at, which Tesla says was 73 MPH.

Tesla finally clarifies fatal Texas crash, confirms driver manually overrode acceleration

Elluswamy also revealed that Tesla data showed Butler overrode the system by pressing the accelerator to 100%, and that the pedal was compressed fully even after the car had crashed. Tesla has not released this data to the public, likely because it is communicating with agencies like the NHTSA on an investigation.

The suit uses a Washington Post analysis of government data that “identified at least 17 fatal incidents linked to Tesla Autopilot.”

This is far from the first time an accident has been blamed on Autopilot. A fatal crash in Texas was blamed on Autopilot several years ago, but when Tesla released data to the NTSB, which was investigating the crash, Autopilot was not available where the crash occurred, and Autosteer was never enabled, meaning the car was manually controlled at the time of the accident.

More information on the accident will be released as Tesla works with agencies to find the cause of the crash. From personal experience, it is hard to imagine Tesla Autopilot or FSD operating in this manner. It drives sometimes too cautiously in residential areas in parking lots, at least in my experience. Speeding happens, but at this rate in this type of area, it is hard to believe.

We look forward to more details being released with time.

Continue Reading

Cybertruck

Tesla Cybertruck is officially the safest pickup, IIHS says

Published

on

Credit: Tesla

The Insurance Institute for Highway Safety (IIHS) has awarded the 2025-2026 Tesla Cybertruck crew cab pickup its highest honor: Top Safety Pick+. This marks the Cybertruck as the only full-size pickup to achieve this distinction in recent evaluations.

The award applies specifically to vehicles built after April 2025, following structural upgrades including front underbody reinforcements and footwell modifications.

These changes enabled strong performance in updated crash tests. The Cybertruck earned “Good” ratings in the small overlap front (driver and passenger sides), updated moderate overlap front, and updated side tests—core requirements for the Top Safety Pick+ designation.

It also secured acceptable or good headlights across trims and a “Good” rating for its standard front crash prevention system in pedestrian scenarios, along with acceptable or good performance in vehicle-to-vehicle testing.

The Cybertruck avoided every single pedestrian collision, including:

  • Daytime child crossing
  • Nightitime adult crossing
  • Night parallel adult

In the large pickup category, competitors such as the Toyota Tundra received only a standard Top Safety Pick, while the Ford F-150 and Ram 1500 did not qualify for either award. This positions the Cybertruck as a standout in occupant protection and crash avoidance among its peers.

Credit: IIHS

Ironically, the same vehicle celebrated for superior U.S. safety performance remains banned from public roads in the United Kingdom and much of Europe. Regulators there cite the Cybertruck’s sharp external edges and highly rigid stainless-steel construction as failing pedestrian-protection standards. European and UK rules require rounded surfaces on protruding parts to minimize injury risk in collisions with vulnerable road users.

Critics also point to the truck’s substantial weight and unyielding body structure, which some argue could transfer more force to other vehicles or pedestrians rather than absorbing it.

Tesla’s engineering philosophy underpins the Cybertruck’s strong IIHS results. The vehicle features a distinctive stainless-steel exoskeleton made from ultra-hard 30X cold-rolled stainless steel. This provides exceptional structural rigidity and a robust safety cage that resists deformation in side impacts and rollovers.

Engineers designed integrated load paths to channel crash forces away from the occupant compartment while allowing controlled energy absorption in key zones. Post-April 2025 refinements to the front underbody further optimized performance in overlap crashes.

Complementing the passive structure is Tesla’s advanced active safety suite, including the standard Collision Avoidance Assist system with automatic emergency braking. This contributed directly to the vehicle’s strong front crash prevention scores. The skateboard platform and low center of gravity also enhance stability and handling, reducing the likelihood of certain crashes.

The IIHS recognition highlights how Tesla’s combination of high-strength materials, structural innovation, and software-driven safety systems can deliver top-tier protection in rigorous testing. While global regulatory differences on design and pedestrian interaction continue to limit the Cybertruck’s availability outside North America, its U.S. safety credentials set a new benchmark for full-size pickups.

Continue Reading

Elon Musk

SpaceX’s newest Starmind will make earth data centers obsolete

Elon Musk confirmed Starmind as SpaceX’s AI satellite constellation name, targeting one million orbital compute nodes.

Published

on

By

Elon Musk confirmed that Starmind will be the official name of SpaceX’s planned AI satellite constellation, following a trademark filing by xAI that surfaced earlier this week. Starmind is what’s being described to the FCC as a constellation of up to one million AI satellites

It’s worth noting that SpaceX’s Starlink communication satellite and Starmind are built on the same orbital infrastructure concept but serve entirely different purposes. Starlink is a connectivity network, with satellites receiving and relaying data between points on Earth, and functioning as a high-speed internet backbone in space. The satellites themselves do not process or think, and move information from one place to another, the same function a fiber cable performs underground.

SpaceX just forced Verizon, AT&T and T-Mobile to team up for the first time in history

Starmind, on the other hand, is something completely different, and tather than moving data, its satellites would compute data through artificial intelligence and directly in orbit using onboard processors powered by large solar arrays. Where a Starlink satellite is essentially a very fast pipe, a Starmind satellite is a server. The practical implication is that Starmind would allow AI models to run inference, process queries, and generate outputs from space, then beam results down to users anywhere on Earth within milliseconds, and without the data ever needing to travel to a terrestrial data center.

Starship will be able to carry 30 to 50 AI1 satellites per launch, delivering the equivalent of dozens of server racks per flight, with no land acquisition, no power grid approval, and no cooling infrastructure required on the ground.

SpaceX is pursuing this new technology as terrestrial data centers are running into hard limits such as lack of physical space, community opposition, and power and water consumption at a scale that is increasingly difficult to permit. Space has unlimited solar power, natural vacuum cooling, and no zoning boards. Musk said in a June 8 video presentation that he expects space to become the lowest-cost location to deploy AI compute within two to three years. Two AI1 prototypes are scheduled to launch in early 2027, with volume production targeted for the end of that year at a new facility called Gigasat.

The real world applications Starmind enables extend well beyond powering Grok. A constellation of orbiting AI processors could run inference workloads for any paying customer, anywhere on Earth, with latency measured in milliseconds rather than the seconds associated with ground-based cloud routing across continents. Starmind, if it scales as described, would make SpaceX the landlord of AI compute the same way Starlink made it the landlord of satellite internet.

Continue Reading