Connect with us

News

Mercedes gets approval for Level 4 self-driving tests in China

Credit: Mercedes

Published

on

Mercedes-Benz has received approval in China to test Level 4 autonomous vehicles on certain roadways, ahead of Tesla’s plans to launch its Full Self-Driving (FSD) this year.

The automaker announced the approval from officials in Beijing over the weekend, letting them test Level 4 autonomous driving in which the vehicles “do the driving for most scenarios without human takeover,” as detailed in a report from South China Morning Post. The testing will focus on driving maneuvers, including U-turns, changing lanes, parking, roundabouts, toll booths, and more.

The approval marks the first foreign automaker to gain approval, and it comes as Tesla expects to gain full approval for its Full Self-Driving (FSD) by the end of this year.

RELATED:

Advertisement

Mercedes will take on Tesla’s FSD in China

Mercedes also gained approval to test Level 3 autonomous driving in China in December, and it says it’s implementing a “minimal-risk strategy” in all of its operations of fully driverless vehicles.

During Tesla’s Q2 2024 earnings call last month, Elon Musk said that the company was likely to gain regulatory approval for FSD Supervised in Europe and China by the end of the year, though its capabilities would be far behind those of North America at first. Tesla has gained a wave of tentative approvals in China throughout this year, and reports in June suggested that the company would be testing 10 vehicles with FSD software ahead of a wider rollout.

Tesla’s FSD Supervised is also generally considered to be a Level 2 of autonomy since it requires driver supervision at all times, according to the SAE International driving automation standards. It’s worth noting that Tesla’s FSD Supervised is available to consumers and is constantly being tested and trained by every driver who engages the system, which is eventually expected to lead to higher levels of autonomy.

Advertisement

Credit: SAE International

According to the SAE, Level 3 autonomous or automated driving includes cases where the driver does not have to monitor the vehicle’s driving, though they may have to take over and drive when the feature requests. Level 4 and 5 autonomy are categorized as those where driver interaction isn’t required at all.

Local companies such as Baidu, Hongqi, and WeRide have also been approved to test Level 4 autonomy, though Mercedes is the first from another country to gain approval. Another self-driving hopeful includes a joint venture between General Motors (GM) and Chinese company SAIC, which launched its own Level 2 autonomous driving software last month.

Mercedes approved for turquoise lights to signify self-driving mode

What are your thoughts? Let me know at zach@teslarati.com, find me on X at @zacharyvisconti, or send us tips at tips@teslarati.com.

Advertisement

Zach is a renewable energy reporter who has been covering electric vehicles since 2020. He grew up in Fremont, California, and he currently lives in Colorado. His work has appeared in the Chicago Tribune, KRON4 San Francisco, FOX31 Denver, InsideEVs, CleanTechnica, and many other publications. When he isn't covering Tesla or other EV companies, you can find him writing and performing music, drinking a good cup of coffee, or hanging out with his cats, Banks and Freddie. Reach out at zach@teslarati.com, find him on X at @zacharyvisconti, or send us tips at tips@teslarati.com.

Advertisement
Comments

Investor's Corner

Tesla has its answer to auto growth, it just has to bring it to the U.S.: analyst

Published

on

Credit: Tesla China

Tesla has its answer to grow its automotive sales over the next few years, TD Cowen analyst Itay Michaeli says, but it just has to bring it to the U.S.

On Thursday, Michaeli reiterated his $490 price target and the ‘Buy’ rating he already held on Tesla stock (NASDAQ: TSLA). However, its automotive division has struggled to show sequential growth over the past few years, mostly due to its focus on AI and Full Self-Driving. Tesla already axed two of its lower-volume vehicles with the Model S and Model X earlier this year.

However, Tesla does not need to engineer an entire new vehicle to trigger an upward tick in sales; it just has to bring it from China to the U.S., Michaeli said.

He is talking about the Model Y L, a slightly larger version of the all-electric crossover that is already available in China. U.S. customers have been pleading with CEO Elon Musk to bring it to the country since its launch in Asia last year, but he’s not convinced of it because of the advent of self-driving and its importance in this particular market.

Advertisement

The problem is that Tesla owners have been requesting something larger that could fit a typical American family. The Model Y L is slightly larger than the standard Model Y, but some are concerned that it could still be too small to fit what most people might need.

Instead, they have asked for a full-size SUV from Tesla.

Tesla gives big hint that it will build Cyber SUV, smaller Cybertruck

Nevertheless, the Model Y L still presents a great opportunity for Tesla in the U.S., and Michaeli says that there is an additional sales opportunity of about 100,000 units, with demand potential falling somewhere between 60,000 and 135,000 units.

Advertisement

TD Cowen’s note to investors also analyzed that Tesla’s growth could come from a stock perspective as well, positively impacting the stock price, as it has been widely reliant on vehicle sales, even though Tesla has truly phased itself away from that being an important metric.

Tesla stands to gain greatly from the introduction of the Model Y L in the U.S., but only if Elon Musk sees it as a viable fit for the market. Families may need to see Tesla bring something larger to the U.S., or they might be forced to buy from another automaker that offers something that fits is needs for more interior space to haul around the kids.

Continue Reading

Elon Musk

Tesla Hardware 3 owners could be made whole this month

Published

on

tesla-asia-model-3
Credit: Tesla Asia/Twitter

Tesla Hardware 3 owners are set to get a new Full Self-Driving version this month as the company plans to release what it is referring to as v14 Lite.

The rollout is not yet confirmed for June, but Tesla executives have stated on several occasions that this more refined FSD iteration will work with their cars and increase its capabilities.

This comes after Tesla admitted during its last Earnings Call that these Hardware 3 vehicles would not be able to achieve Full Self-Driving, something that they did not know when they bought these cars. We regularly receive messages from Hardware 3 owners asking when v14 Lite will come out, what they should expect, and whether it is worth it to upgrade the self-driving computer or buy a new car altogether.

It is hard not to feel for them; Tesla CEO Elon Musk said at the company’s 2019 Autonomy Day that all vehicles produced at the time, including Hardware 3 cars, had “all the hardware necessary, compute and otherwise, for Full Self-Driving.”

Advertisement

Musk also said in March of that year that, “Anyone who purchased Full Self-Driving will get FSD computer upgrade for free.”

However, during the Q1 2026 Earnings Call, Musk admitted that Hardware 3 vehicles would not be capable of FSD, as “It has only 1/8th the memory bandwidth of Hardware 4, and memory bandwidth is one of the key elements needed for unsupervised FSD.”

Advertisement

Tesla has made some effort to remedy these Hardware 3 owners by offering:

  • Discounted trade-ins toward AI4 cars
  • Hardware retrofits, which would replace the self-driving computer and upgrade all cameras
  • Full Self-Driving v14 Lite

The issue is that many of these owners were led to believe their cars would be capable of unsupervised self-driving. Now, they’re left scrambling for options, and while there are several, they will all require more money out of their pockets.

Expectations for Tesla v14 Lite for Hardware 3 Owners

The big differences between the AI4 v14 and v14 Lite for Hardware 3 owners will stem primarily from hardware constraints. Tesla developed v14 Lite with an optimized frame of mind; the v14 neural nets are toned down to run on an HW3 computer.

Tesla v14 will use the same behavior, but its limits will be hardware-related, especially given that the cameras on HW3 vehicles are lower-resolution.

Tesla reveals its plans for Hardware 3 owners who are eager for updates

Advertisement

This will result in potentially more edge cases due to the lower quality perception and less long-range detection, but reaction time and overall confidence should be more refined.

There should also be a handful of additional features that are available on AI4 cars, such as:

  • Starting Full Self-Driving from Park
  • Auto Shift
  • Streaks
  • Speed Profiles
  • Improved Dynamics, like Pulling Over for Emergency Vehicles

Tesla plans to release v14 Lite this month, but we are all familiar with how the company can be with timelines. Additionally, if v14 Lite has not proven to be ready for a wide release, Tesla will slam the brakes on the rollout.

We would anticipate that Tesla is testing v14 Lite internally, and likely has been for several months.

Advertisement
Continue Reading

Elon Musk

SpaceXAI just launched into your kitchen with their new app

SpaceXAI just powered its first consumer app and it predicts what you want to buy.

Published

on

By

SpaceXAI just made its first move into consumer AI, and it involves your grocery cart. On June 3, 2026, Gopuff and SpaceXAI announced the launch of Go, a Grok-powered shopping assistant built directly into the Gopuff app that predicts what you need before you even start searching for it.

Gopuff is an instant delivery platform that operates more than 400 micro-fulfillment centers across the U.S., delivering everyday essentials, snacks, drinks, and household items in as little as 15 minutes. It is not a restaurant delivery app or a marketplace. It owns its inventory, controls its warehouses, and handles its own logistics, which means it has built one of the most detailed consumer behavior datasets in retail over its 13-year history.

Go combines SpaceXAI’s advanced reasoning, voice, and image generation models with Gopuff’s dataset of hundreds of millions of orders and real-time cultural signals from X to prepare a suggested cart the moment a customer opens the app. It learns each shopper’s habits and automatically builds a personalized cart based on time of day, location, order history, and real-time indicators. Returning customers can check out with a single tap.


Rather than searching for specific items, users can describe a situation like a game-day party or the desire for a healthy breakfast and Go will assemble a cart automatically. It can also predict when shoppers are running low on items like coffee or paper towels and have them packed and delivered in under 15 minutes. Grok voice integration lets users talk to the app in plain conversational language and check out completely hands-free.

Advertisement

Gopuff co-founder and co-CEO Yakir Gola said: “Today, we believe the greatest friction left in commerce is not delivery or instantaneous access to the essentials customers need. It’s the moment before: the thinking, the deciding, the remembering. We’re combining Gopuff’s demand intelligence with xAI’s frontier reasoning to create an everyday shopping experience that feels like a true extension of you.”

Why SpaceX just made a $60 billion bet on AI coding ahead of historic IPO

The timing carries context beyond the product launch. SpaceXAI was formed after SpaceX completed an all-stock merger with Elon Musk’s xAI earlier this year, folding one of the most advanced AI labs in the world into the same corporate structure as the company preparing what could be the largest IPO in history. SpaceXAI is dipping into consumer-focused AI just as it prepares for its public debut, and while Musk has openly discussed building an everything app, this launch uses Grok to power another company’s product rather than launching a standalone consumer platform. Every consumer-facing deployment of Grok ahead of the IPO roadshow adds tangible evidence that SpaceXAI is not just an infrastructure play but a direct competitor in the AI application layer where OpenAI and Google are already fighting for dominance.

Advertisement
Continue Reading