Large language models gave artificial intelligence a working recipe. Pretrain a large model on broad data, and general capability follows. Robotics has no such recipe. Robotics systems have long been assembled from separate perception, planning, and control parts that rarely add up to intelligence a robot can carry from one task to another, or one machine to another. The central problem in embodied AI is to find the equivalent recipe, and the field does not yet agree on what it is.
X Square Robot, a Chinese embodied-AI company, has made an unusually explicit bet. It argues that the recipe is an integrated stack, spanning the data a robot learns from, a world model for predicting changes in the physical world, and an action model that brings together perception, planning, reasoning, and decision-making to generate executable robot behavior. The company also believes that the stack should be built and released in the open.
X Square Robot shares its vision of bringing robots into real homes.X Square Robot
Advertisement
X Square Robot’s embodied AI stack
What holds the stack together is a small set of principles rather than a single overarching model.
The first is that the basic unit of robot data is an interaction, not a trajectory; a demonstration is successful only if it changes the world as intended, not simply because the joints moved.
The second is that pretraining should yield usable capability, not just an initialization for later fine-tuning.
The third is that behavior should be modeled around physical events rather than fixed slices of time.
These principles make the layers interdependent, since the same robot-free data that trains the action model is also structured to feed the world model. It is worth being precise, though. The company describes the world model and the action model as complementary but independent model families that share a code base. Both sit within its broader World Unified Model, which it has presented as an architecture for training vision, language, action, and physical prediction together.
Robot learning data: Engineering for quality and cost, not scale
For the X Square Robot team, one of the biggest constraints on general-purpose robots is the cost and quality of interaction data, not the number of parameters. To address that, the company built its Universal Manipulation Interface (UMI) data collection system, QUANXTA Zero Series. It works by collecting demonstrations from people wearing a rig with dual grippers rather than teleoperating a robot. This approach is not itself new, and builds on established methods for robot-free data capture. What sets it apart are two engineering choices.
The first is quality control, and it is the most distinctive part. Rather than accepting recorded trajectories as they are, the system runs a closed inspection loop, and its notable step is physical playback. A sample of trajectories is replayed on the real robot, and only those that actually complete the task count as valid. That makes the validity rate a measured quantity rather than an assumption. For example, a gripper that closes a fraction of a second too early still looks like a grasp in the data, yet it has pushed the object away, so it shouldn’t be classified as valid. A smaller clean dataset can be worth more than a larger noisy one.
The second choice is how lower-cost human data and scarce robot data are combined. The company pretrains on a large volume of robot-free demonstrations to build general representations, then adds a small amount of real-robot data as an anchor to the specific machine’s dynamics. It reports that this reaches performance comparable to an all-robot dataset at roughly a 20-fold lower cost of collection, driven mainly by how much cheaper the wearable rig is than a teleoperation setup.
Advertisement
The resulting dataset is deliberately model-agnostic, formatted to feed both action models and world models. The caveat is that the strongest results are measured on the company’s own robots and data-collection pipelines. Broader independent testing will help confirm and extend these promising results across a wider range of settings.
A world model organized around events
In developing its world model, called WALL-WM, X Square Robot took a differentiated approach. Most action models predict a fixed-length chunk of motion from the current image and instruction. That is convenient, but it segments behavior into fixed-duration windows, so the boundaries fall where elapsed time dictates rather than where one action ends and the next begins. WALL-WM instead treats an action-grounded semantic event as its unit: a coherent piece of behavior such as reaching, grasping, or placing, something that can be named in language, seen in video, and executed as motion.
WALL-WM’s design reflects a specific concern about not discarding what large video models already know. To achieve that, a text-to-video model is coupled to a freshly initialized action network that reads from the video features without overwriting them, which preserves the visual prior. From that one process, it offers two modes. An event mode runs in variable-length segments and suits reasoning over long horizons, while a fixed-length mode produces the steady, real-time output a controller needs. That places WALL-WM between mainstream chunk-based action models and pure video world models, keeping the predictive character of a world model while still yielding executable control.
In a series of experiments, the company relied on a generalization test that is more specific than most. A model trained on a limited dataset was evaluated on long-horizon tasks in unseen settings and, on the company’s real-robot benchmark, reportedly outscored baselines that had been fine-tuned on related data. That is a meaningful result if it holds. For now, it is measured on the company’s own benchmark. With the code now being released, the broader community will have the opportunity to test, reproduce, and build on them across more settings.
Advertisement
A policy that runs before fine-tuning, and action tokens with meaning
The action layer carries two connected ideas. The first is a requirement the company sets for itself with Wall-OSS-0.5, its vision-language-action model: The pretrained model should run on a real robot before any task-specific fine-tuning.
The interest is less in the scores than in the design behind them. The model trains three objectives together, namely discrete action tokens, language grounding, and continuous action generation. And it keeps gradients flowing through all of them rather than freezing parts of the network as some rival designs do. It’s also a more strict method, since it reports untuned behavior such as approaching, grasping, and recovering, including on a deformable task held out of training.
The second idea is the action interface itself, called X-Tokenizer. Most systems that turn continuous motion into discrete tokens produce codes that the language model cannot interpret. X-Tokenizer reframes tokenization as learning a semantic interface, so that the top-level code stands for the intent of a motion while lower-level codes carry finer detail, all aligned with the language model’s own features.
A useful consequence is stability. Adding noise to an action barely moves the intent code, which is what lets one tokenizer to be reused across robots without re-tuning. The tokenizer inside the production action model is a related variant of this approach. Together, the two ideas give the action layer something rather powerful: capability that transfers.
Advertisement
The future of embodied AI stacks
X Square Robot is betting that its unique approach combining three layers, each specialized in solving a key part of the problem, will stand out from other embodied AI stacks. The physical-playback step that grounds data quality is uncommon and sensible. The reframing of world modeling around events, with one backbone serving both reasoning and control, is a genuinely distinct approach. And the pairing of a deployable pretraining standard with a tokenizer designed as a semantic interface gives the action layer unusual coherence.
X Square Robot’s valuation has climbed above 20 billion yuan (about US $2.9 billion), suggesting that investors increasingly view data infrastructure, foundation models, and scalable training systems as long-term differentiators in embodied AI.
The next phase will bring broader validation. Much of the current evidence comes from X Square’s own robots and benchmarks. With the world model code now being made public, and as the community begins to test, reproduce, and build on the work, the reported capabilities will be tested across more robots, tasks, and settings.
X Square Robot’s recent funding rounds reflect similar confidence. The company’s valuation has climbed above 20 billion yuan (about US $2.9 billion), suggesting that investors increasingly view data infrastructure, foundation models, and scalable training systems as long-term differentiators in embodied AI.
Advertisement
What’s next for X Square Robot
To learn more about its future plans, the following Q&A with the X Square Robot team further explores the company’s technology, strategy, and vision.
What made now the right moment, technically, to commit to this stack? What recently became possible that wasn’t possible a couple of years ago?
It is not one breakthrough but several trends maturing together. Foundation models gave us a shared representation across vision, language, and action, so we can model what a robot sees, what it is asked to do, and how its actions change the world in one framework, rather than as separate perception, planning, and control modules.
Compute and infrastructure are finally sufficient for large-scale pretraining over long-horizon, multi-embodiment data. Just as importantly, we realized that data, not model size, is the real bottleneck for general robots—what is scarce is diverse, high-quality, reproducible interaction data. And world modeling has become practical. The useful question is no longer how to predict a few seconds of video, but how to understand the ways actions change objects, contacts, and task states. Two years ago these ingredients existed separately. Today they are mature enough to work as one system.
Advertisement
“We realized that data, not model size, is the real bottleneck for general robots—what is scarce is diverse, high-quality, reproducible interaction data. And world modeling has become practical.”
Your data system captures demonstrations with a wearable VR rig and custom grippers rather than teleoperating robots. What was wrong with standard teleoperation?
Teleoperation is built around controlling the robot. It forces the operator to work within the machine’s kinematics, latency, and viewpoint, and the resulting demonstrations are slower, stiffer, and less diverse. We built our system around capturing human skill instead. Manipulation is really about contact, timing, finger coordination, and recovery, not just the path the hand takes, and a wearable rig records those before the behavior is compressed onto one particular robot. It also breaks teleoperation’s expensive scaling law, in which every demonstration needs a robot.
People can generate rich data independently of any robot, and the crucial property is that those demonstrations can still be replayed and executed on a physical robot through the model. Mobility is convenient, but that replay is the real point, because it is what lets the same data be reused across different platforms.
Advertisement
In X Square Robot’s approach, demonstrations can be replayed and executed on a physical robot through the AI model, allowing the same data to be reused across different platforms.X Square Robot
X Square Robot reports that its pipeline has roughly an 85 percent data-validity rate. Why is quality control such an underrated bottleneck?
Because errors in robot data are far more expensive than in language data. A small timing or contact error can change what a demonstration means. If a gripper closes a fraction of a second too early, the motion still looks like a grasp, but physically it has pushed the object away. A dataset that mixes failures and accidental successes teaches ambiguity, not skill, because the real unit is the interaction, not the trajectory.
So we run automated inspection, kinematic checks, and physical replay, where we play a sample of trajectories back on the real robot and count only the ones that actually complete the task. Data quality sets the ceiling on how good a policy can be. In our experience a smaller, cleaner dataset often beats a much larger, noisier one, which is why we treat quality control as part of the model, not a preprocessing afterthought.
The model runs in both “event mode” and “chunk mode.” When does each matter?
Advertisement
Both matter, for different reasons. The physical world changes through events—when contact occurs, a grasp forms, or an object slips—not in fixed-frame windows. Event mode concentrates the model’s attention on those moments, and it matters most for long-horizon tasks, like clearing a table, where progress is a sequence of semantic events rather than a smooth stream. It runs in variable-length segments that follow the task rather than a clock. Chunk mode matters for deployment. Real controllers need a stable, real-time interface, and fixed-length chunks integrate cleanly with existing control systems.
We organize learning around events in the first place because a fixed window can split one motion in half or merge two together, which turns training into short-horizon pattern matching and weakens the model on long tasks. So the world model’s job is to connect event-level understanding, which is where the reasoning happens, with a fixed-length output a real robot can actually run.
Why make “deployable before fine-tuning” the criterion?
Pretraining should produce capability, not just a good starting point. If a model is only useful after heavy fine-tuning, then most of the intelligence still lives in the downstream supervision, not in the foundation model. Deployable before fine-tuning is a more honest test of what pretraining actually learned. A well-pretrained robot should already know how to approach, grasp, move, avoid obstacles, and correct itself. Fine-tuning should adapt it to a specific task or robot, not create the ability from nothing. It is also a practical requirement. A robot in a home or a workplace shouldn’t need a brand-new dataset and a new policy every time the task changes, so a foundation model that already carries general skill, and some ability to recover, is the minimum bar for something genuinely useful in the real world.
Advertisement
What is the most challenging part of cross-embodiment learning?
Robots differ in control frequency, delay, compliance, sensing precision, and contact dynamics, so the same instruction can require different action decompositions and recovery strategies, and a behavior that works on one arm cannot simply be copied to another. Cross-embodiment learning needs an intermediate abstraction, lower than language but higher than joint angles: how you approach an object, how you make contact, how you apply force, and how you recover from a mistake.
When we say cross-embodiment, the main capability we mean is multi-embodiment generalization: transferring across robots, training on many embodiments at once, and adapting to different kinematics. Human-to-robot transfer and other techniques are specific approaches to that goal.
“A robot in a home or workplace shouldn’t need a new dataset and policy every time the task changes. A useful foundation model should already carry general skills and the ability to recover.”
Advertisement
What would you most like to see other researchers attempt to reproduce or stress-test?
Three things, above all. Whether event-level representations really generalize beyond our own datasets, across more tasks, scenes, objects, embodiments, and failure conditions. Whether pretraining stays effective on robots the model never saw during training, or whether its capability is still too tightly coupled to what it has already seen. And whether real-robot evaluation can become a shared language for the field, so that we compare not just success rates but the reasons systems fail, where an instruction was misread, where perception broke down, or where recovery fell short. Robotics has been driven too often by impressive demonstrations, and real progress comes from results that are reproducible and diagnosable.
What capability is still missing before robots become dependable in homes?
Benchmarks measure competence, like whether a model can finish a task. Homes demand reliability, safe and consistent operation over time in a place that changes every day, with objects moving, instructions that are vague, and people interrupting. The missing piece is not a higher one-time success rate: it is robust recovery. A dependable home robot has to know when it is uncertain, when to slow down, when to ask for help, and how to bring the world back to a safe state after it drops something or misunderstands a request.
Advertisement
In a real home, failure recovery matters more than raw success, because the home does not reset itself. Homes also demand careful personalization, learning a household’s routines and preferences over time, with safety and trust as first principles. That combination, not any single skill, separates a capable demonstration from a robot people can live with.
X Square Robot’s approach is that, in a real home, failure recovery matters more than raw success, because the home does not reset itself and it demands careful personalization, with safety and trust as first principles. X Square Robot
How do the open-source components fit into X Square Robot’s World Unified Model direction?
We see these releases as layers of the World Unified Model direction rather than isolated projects. Wall-OSS-0.5, the action model, asks whether an open vision-language-action model can gain directly measurable capability from large-scale pretraining, so it is the capability layer. WALL-WM, the world model, asks how a robot should understand change in the world, shifting from fixed windows to event-level modeling, so it is the representation layer. The data system supplies the interaction data that both of them learn from.
Together they form a loop in which models produce capability, world models organize understanding, and the open-source community drives reproduction and improvement. World Unified Model is the broader architecture those layers support, bringing vision, language, action, and physical prediction together.
Advertisement
We are releasing these pieces openly because embodied intelligence cannot be solved by one organization; it needs many embodiments, many real tasks, and broad feedback, and the long-term goal is a stack that keeps learning and ultimately moves robots from laboratory demonstrations toward reliable everyday use.
Over the weekend, the Los Angeles Police Department announced it would no longer use the 138 mounted surveillance cameras in the city operated by Flock, which were being used to track vehicles by automatically reading their license plates. A few months ago, the Los Angeles Times reported on discontent in LA over Flock potentially sharing information with the US government, including the Immigration and Customs Enforcement agency, although at the time the LAPD praised Flock’s cameras as “tremendous investigative tools.”
In May, a city council member moved to suspend new contracts or agreements with Flock.
Advertisement
Neither the LAPD, the LA City Council nor Flock Safety responded immediately to a request for comment.
The LAPD told news outlets that a sticking point on moving forward with the company was uncertainty about the data that Flock collects from its cameras, including who owns it and how it can be used once it’s collected.
Flock’s statement to outlets, including KTLA, suggested that misperceptions were driving the decision and that it had been working with the LAPD on data privacy protections and limits around data access. “While this latest development comes as a surprise, we remain committed to continuing our active and ongoing conversations with LAPD to find a path forward,” a Flock spokesperson told the TV station.
Cities dropping or reevaluating Flock
A number of US cities have decided to part ways with Flock Safety this year, but even after contracts ended, some continued to have issues with the company.
Dayton accused the company of data-sharing violations, including searched related to immigration enforcement. Evanston issued a cease-and-desist order after discovering that Flock had reinstalled cameras the city had ordered removed.
Other cities that have canceled contracts with Flock include Mountain View and Oakland, California; Knoxville, Tennessee; Flagstaff, Arizona; Cambridge, Massachusetts; San Marcos, Texas; and Redmond, Washington. The website DeFlock is maintaining a running list of city councils that have canceled contracts or rejected bids from Flock.
As DeFlock points out, Flock isn’t the only vendor of software and hardware that performs automatic license plate reading. Others include Axon, Genetec and Motorola Solutions.
We may receive a commission on purchases made from links.
Not having the right tools to handle things that go wrong and need repairing around the house can quickly make a small problem much more stressful, especially when it’s something you know you can handle yourself. Instead of having to make outcalls to get problems fixed, having high-quality, reliable tools on hand can help you save plenty of time and money in the long run. And there are very few brands that fit this bill better than Milwaukee.
Advertisement
Competing with the likes of Ryobi in just about every category relevant to homeowners, Milwaukee’s smaller, more affordable hand tools are often the most well-reviewed across the board. Many of them are designed to be as practical as possible, combining multiple tools into one compact unit or just taking extra care with the simplest tools. Alongside fixing things, Milwaukee has no shortage of options to help put things together to make building and DIY as easy as possible. Here’s a look at five tools that every homeowner should consider investing in.
Advertisement
11-in-1 Multi-Tip Combination Screwdriver
Whether you’re moving into a new home or are just looking to make some additions, there’s a strong chance you’ll be building things. And to do that, a screwdriver will almost certainly be needed. Furniture sold in pieces is often much cheaper than pre-built items, so having the tools to build it all will make a massive difference. Unsurprisingly, Milwaukee offers plenty of screwdrivers, but this 11-in-1 combination screwdriver is useful for putting furniture together, but its use doesn’t end there.
If you feel confident that you can handle various electrical issues that often come up in a house, one of the eight bits included with the driver can take out the bolts used on power outlets and their cases, as well as things like light switches. This screwdriver comes with two Philips, two slotted, two square, and two Torx bits, alongside the three nut drivers, to help cover all bases with your home electronics. While it’s a professional-grade tool, its versatility is what makes it worth the $11.97 it’s sold for at Home Depot, where over 900 user reviews give it an average of 4.5 stars out of five.
Another tool that’s geared towards professional use but will still prove massively useful is the combination electrician’s 6-in-1 Wire Stripper and Cutter Pliers. Similar to the combination screwdriver, this is another Milwaukee piece that prevents you from having to buy multiple tools, ultimately saving you money while being more practical.
If you do have the skills, there isn’t much in regard to small electrical installations and fixes that this tool won’t be able to help out with. The wire stripping element has the ability to strip solid wire between 8 and 18 AWG for solid wire, and between 10 and 20 AWG stranded wire. The wire cutter itself has a slight curve to it, making it more adept at making clean cuts on stranded pieces as well as the standard solid wire. The tip of the pliers doubles as a reamer and a regular grooved plier head, ideal for gripping and shaping. The price of $19.97 on Home Depot is pretty good for how much you can do with this tool.
Advertisement
25 ft. Magnetic Tape Measure
This next product from Milwaukee is a little more standard of a tool than others here, and there’s a good chance you already have a few lying around the house. However, if you ever decide to renovate your home, whether that’s something major like a new kitchen or you just want to move things around, you’ll quickly appreciate how useful a reliable tape measure can be. Milwaukee again has many different options, but based on reviews, this 25-foot magnetic tape measure more than stands out.
While having a magnetic tape measure isn’t essential, having the ability to easily latch onto hard-to-reach places could easily come in handy at some point down the road. The blade itself falls in line with a few other options from Milwaukee, having a maximum reach of 15 feet. The more important rating is the 12 feet of standout, though, making this a great option for a one-person job. Owners note how sturdy the tape measure is. At Home Depot, this Milwaukee magnetic tape measure goes for around $25.
Advertisement
Adjustable wrench
While a nut driver can easily handle smaller external fasteners that are common throughout most homes, they don’t always provide enough leverage for more demanding jobs. Whether it’s building or maintenance work, having a solid wrench on hand will never be an inconvenience. Similar to the tape measures, Milwaukee offers plenty of different sizes for its adjustable wrench, ranging from six inches up to 15 inches, for when you’re working with extra stubborn fasteners. The smaller sizes are affordable, but the larger ones do climb up the price ladder quite quickly.
Advertisement
This is an inherently simple tool compared to others, even on this list, doing one job and one job only. Milwaukee still makes sure all areas are optimized, however, no matter what size you go for. Measurements are lasered on to create a ruler below the jaws. The screw itself uses a proprietary system to keep the jaws firmly locked onto fasteners, and the smooth, slightly curved handle is designed to be as comfortable as a wrench can be. Reviews of all adjustable wrench sizes from Home Depot confirm how effective these tools are, with the bundles of different sizes being quite a popular choice. Going down the middle for an eight or 10-inch wrench should be more than enough for tightening external bolts around the house, though.
Advertisement
Fastback 6-in-1 Folding Utility Knife
If you want a tool that can be applied and be useful in so many different jobs around the house, Milwaukee’s Fastback Folding Utility Knife can be one of the best time-savers you can buy from this brand, for a few different reasons. This specific knife comes with a general-purpose blade pre-installed, but it also has the ability to hold different compatible blades if you want to buy them as well. But even with the standard blade that extends 1.27 inches from the body can help you with all sorts of DIY projects and repairs. Simple tasks like opening well-packaged boxes also become a breeze.
While having a high-quality blade is worth it alone, this Milwaukee knife has other features that help it earn a $21.97 price tag at Home Depot. You get a small wire stripper built into the blade guard, making it even more useful for electrical work. The folding screwdriver is also a neat addition, and it’s always there in case you need it. It comes with a reversible Philips #2 bit and a slotted 1/4-inch bit. You also get a built-in bottle opener, saving you from needing another item with you, even for something as minor as this. This is unsurprisingly one of the most reviewed Milwaukee products we included on this list, with almost 2,000 reviews averaging 4.4 stars out of five.
Advertisement
Methodology
When selecting Milwaukee tools for this list, we first made sure that they were readily available from The Home Depot, the biggest hardware store that sells Milwaukee tools. Then, we checked that each tool had at least 100 reviews averaging a score of at least four of out five stars. We only selected tools geared towards general DIY and repair jobs that can come up in any home, nothing too specialized, and made sure to find some user reviews mentioning how useful they can be around the house.
In a secluded room deep within Samsung Display’s headquarters in South Korea, rows of whirring gray and black machines repeatedly fold, flex and stress test the company’s newest mobile displays. During a mid-June visit, I was among the first people outside the company to step inside the high-security lab and see how Samsung pushes its foldable screens to their limits before they reach consumers.
On Tuesday, Samsung unveiled Flex Titanium, a new display technology for its upcoming Galaxy foldable phones including the Z Fold 8. It combines a titanium-alloy film with a titanium plate to create a thinner, more durable display structure designed to better withstand drops and other impacts — an important consideration for foldable phones that can cost thousands of dollars.
Samsung Display designs and manufactures screens for Samsung Electronics as well as competitors including Apple, and has become one of the industry’s leading developers of flexible and advanced display technology. Beyond commercial products, the company regularly showcases futuristic display concepts for phones, tablets and other devices.
Watch this: I Went Inside Samsung’s Secret Display Lab and Saw Its Wildest Phone Concepts
As Samsung makes its foldable phones thinner — last year’s Galaxy Z Fold 7 measures an impressively slim 4.2mm when open — the company is looking for ways to scale back various components to maintain a sleek profile. Samsung says it spent about three years developing Flex Titanium technology while examining customer feedback across seven generations of its foldable phones.
“We have to understand user behavior and various display challenges like dropping or pressure with a large object or a tiny object,” Samsung executive vice president Byung Duk Yang said in an interview. “Because of that, we have developed a very comprehensive and sophisticated evaluation method to understand user behavior in the real world.”
Advertisement
These are the machines that fold Samsung’s displays hundreds of thousands of times to ensure durability.
Samsung
Testing the endurance of foldable displays
As we navigated the maze-like, pristine white hallways snaking below Samsung Display’s headquarters in Korea, about 20 miles from Seoul, our guide touted the exclusivity of what we’d be seeing. No one outside the company — not even the employee’s mom and dad — had been here, she said as she led us into the testing lab.
In this secluded room, which only engineers enter, equipment runs around the clock, folding and unfolding display panels to ensure they can pass 500,000 folding tests. Once the metal latch is closed, the Z Fold 8 panels (there are four inside right now) are subjected to extreme temperatures ranging from -20 degrees to 60 degrees Celsius (or -4 degrees to 140 degrees Fahrenheit).
Advertisement
Outside the machine, a monitor shows what’s happening inside from eight different camera angles. The footage, which is also being recorded, can detect issues such as the display lifting off the device frame. Currently, the display panels being tested are off, but the machine can evaluate how operational displays respond to extreme conditions, too.
The machine to my left tests a display’s image quality. You can slide open the panels to peer inside via the small windows.
Samsung
Down another long hallway (I feel like I’m in an episode of Severance at this point), we enter a lab for examining the display’s image quality, including brightness and color. After placing a display in the center of the machine and closing what looks like a miniature garage door with sliders on the windows for peering in, the testing begins.
Advertisement
A series of bright lights beams down on the panel, and the machine measures how much light is reflected — the less, the better. That’s for a few reasons: The display’s colors appear deeper, less reflection makes it easier to see the screen under bright lighting and you’re less likely to just be staring at your reflection when you look down at your phone. It takes about three minutes to test one panel.
The ball drop test ensures a display can effectively absorb and distribute pressure from a small, heavy metal ball.
Samsung
Another test I saw appeared rather simple compared to these more deeply technical mechanics, but it’s equally important for ensuring a display’s durability. A towering 220-pound machine propped on a counter holds a marble-sized metal ball weighing around 21 grams. An arm-like structure drops the ball from a height of 30 centimeters onto the display three times to ensure it won’t crack. In our demo, we pushed the limits to higher drops from 40 and 50 centimeters, and the display effectively absorbed and distributed the pressure to avoid damage.
Advertisement
This marble-sized ball is small but mighty. It’s dropped onto a display panel to test if it cracks.
Samsung
Making a “better display”
Samsung’s new Flex Titanium is all about bolstering durabily without adding bulk. Compared to polymer film, titanium-alloy film has 20 times greater mechanical stiffness, the company says, meaning it better retains its shape. This component sits below the OLED panel and is less than one-third the thickness of a human hair. Below that is the titanium plate, which Samsung says can provide more stable support when the phone is unfolded without compromising flexibility.
Last year’s Z Fold 7 also used a titanium plate, but the display structure was made up of multiple polymer-based support layers. Samsung has now consolidated those layers into a single titanium-alloy film, reducing the thickness of the display module while maintaining strength, flexibility and long-term durability, the company says.
Notably, the upgraded display also has a reduced crease — a growing focal point in the world of foldable phones. Samsung Display showed off a creaseless foldable concept screen at CES earlier this year. The company is reportedly working with Apple to develop a creaseless screen for a foldable iPhone, which could make its debut this fall. What I saw at Samsung Display in June still had a minimal crease, although it’s much less apparent than the Z Fold 7’s.
Advertisement
At CES 2026, Samsung Display showed off a concept for a creaseless screen.
Celso Bulgatti/CNET
Samsung is slated to share more about Flex Titanium and upcoming Galaxy foldable devices, including the Z Fold 8, during its summer Unpacked event on July 22. These advancements look to be a step toward mitigating many of the durability and aesthetic compromises that have long characterized foldable phones — though the work is far from done.
“Years ago, Samsung created this foldable category,” Yang said. “And the foldable display and the structure we developed became the standard. So we feel some responsibility for this market; we have to make a better display.”
Twenty-six anonymous Meta employees have sued the company in federal court in Oakland, California, alleging it used AI-powered software to disproportionately target workers with disabilities, medical leave histories, and pregnancies during mass layoffs.
The employees are seeking to halt the next round of layoffs, scheduled to begin on Jul 22, while they pursue arbitration.
The lawsuit, filed on July 13, said that this disadvantages people who missed work because of medical conditions or to care for family members, violating federal and state discrimination and retaliation laws.
Advertisement
However, Meta has refuted the claims, stating that they lack merit and that people, not AI, made the workforce decisions.
The lawsuit appears to be the first against a major company to challenge the alleged use of AI in conducting layoffs.
Read other articles we’ve written on Singapore’s job landscape here.
HWMonitor is a lightweight hardware monitoring program that reads your system’s main health sensors. It provides real-time data on voltages, temperatures, fan speeds, and power consumption for CPUs, GPUs, motherboards, storage devices, and more.
Whether you’re troubleshooting, stress testing, or just keeping an eye on system health, HWMonitor delivers clear and reliable metrics in a simple interface. Ideal for PC enthusiasts, overclockers, and tech professionals. The program handles the most common sensor chips, like ITE IT87 series, most Winbond ICs, and others. In addition, it can read modern CPUs on-die core thermal sensors, as well has hard drives temperature via S.M.A.R.T, and GPU temperature.
Why does HWMonitor show several different CPU temperatures?
HWMonitor lists core temperatures individually (Core 0, Core 1, etc.) and also shows an overall CPU package temperature. The core readings are from thermal sensors inside each CPU core, while the package temperature represents the hottest part of the CPU die. When monitoring for thermal issues or stress testing, the highest of these values is the most important.
Is HWMonitor accurate for reporting GPU temperatures?
HWMonitor is generally accurate for GPU temperature readings, as it pulls data directly from the onboard sensors. However, some users report small differences when comparing with MSI Afterburner and other tools. If you need exact numbers for overclocking or thermal analysis, it’s worth comparing with a second tool.
Advertisement
Can HWMonitor show how much power my GPU is using?
Yes, HWMonitor can display GPU power usage if your graphics card supports it. It shows readings from the PCIe slot and the auxiliary power connectors (6-pin or 8-pin). By adding these values together, you can estimate the total power draw of your GPU. This can be helpful when deciding whether your power supply is sufficient or if you’re planning an upgrade.
What are TMPIN readings in HWMonitor?
TMPIN labels are temperature sensors on the motherboard, but their exact purpose can vary depending on the manufacturer. They might monitor the VRMs, CPU socket area, chipset, or other components. Because the naming isn’t standardized, the readings can be hard to interpret without checking your motherboard documentation.
Which CPU temperature should I focus on: package, cores, or motherboard?
The most important temperatures to monitor are the CPU package and the highest individual core temperatures. These reflect the real thermal state of your processor. Motherboard CPU readings are often lower and less precise. As long as your CPU stays below 90 – 95 °C under load, it’s within safe limits for most modern CPUs.
What’s New:
Hotspot temperature on NVIDIA RTX 50×0 GPUs.
Preliminary support of Lisuan 7G100 GPU.
Previous Release Notes:
Intel Arc G3 & G3 Extreme (Panther Lake).
Intel Core Ultra 5 250KF Plus (Arrow Lake Refresh).
AMD Ryzen 7 7700X3D (Raphael).
AMD Ryzen AI Max+ 495, 492, 488 (Gorgon Halo).
AMD Ryzen AI Max 490, 485 (Gorgon Halo).
AMD Ryzen AI Max PRO 495, 490, 485, 480 (Gorgon Halo).
AMD Ryzen 9 9950X3D2 (Granite Ridge).
AMD Ryzen 9 PRO 9965X3D, PRO 9945 (Granite Ridge).
1Password on Tuesday launched AI Spend and Consumption Management, a new capability embedded in its SaaS Manager platform that gives IT and finance teams a unified, real-time view of how their organizations consume and spend on AI services from vendors including Anthropic, Cursor, and OpenAI.
The move marks the latest strategic expansion for a company that built its reputation on password management for consumers and, over the past three years, has aggressively repositioned itself as a broader identity security and SaaS governance platform for enterprise buyers. With this release, 1Password is staking a claim in one of enterprise technology’s newest and most chaotic budget categories: the consumption-based cost of large language models.
“Executives want teams to build faster with AI, but that speed is creating a new kind of spending pressure,” Greg Henry, 1Password’s chief financial officer, said in an exclusive interview with VentureBeat. “Developers are consuming tokens at a pace that traditional budgets weren’t built to manage, and IT and finance teams are being asked to forecast and justify AI investments without a clear view of what’s actually driving costs.”
The product, now in public preview with broad availability planned for fall 2026, connects directly to vendor admin APIs to pull token-level consumption data daily. It normalizes that data across providers into a single dashboard and allows organizations to set vendor-level spend limits, configure threshold-based alerts via Slack and email, and break down usage by team, user, vendor, and model.
Advertisement
Why traditional software budgets can’t keep up with AI token pricing
The core challenge 1Password is targeting is structural. Traditional SaaS pricing operates on a per-seat, per-year model that is easy to budget and reconcile. AI pricing does not. Every API call to Claude, GPT-5.6, or a Cursor-powered coding assistant consumes tokens, and the cost of those tokens varies by model, by input versus output, and by the complexity of the task. A single engineering team running agentic workflows can burn through a prepaid token budget in weeks — and the finance team may not notice until the invoice arrives.
Henry drew a sharp analogy to a problem enterprises have already lived through once. “Consumption-based pricing isn’t new,” he said. “We saw it arrive with cloud infrastructure, and it took years to build the tools and disciplines to manage it. AI is the next version of that shift.”
That comparison resonates across the industry. When Amazon Web Services, Microsoft Azure, and Google Cloud popularized consumption-based pricing for compute and storage in the 2010s, enterprises initially lacked the tooling to monitor and optimize their cloud bills. That gap spawned an entire FinOps ecosystem — companies like CloudHealth, Spot.io, and Apptio built multi-billion-dollar businesses helping organizations understand what they were spending on cloud and why. Henry is explicitly betting that AI token spend will follow the same trajectory, and that organizations that fail to build visibility now will end up, as he put it, “paying far more than they needed to, for far longer than they should have.”
Advertisement
The scale of the coming wave lends credibility to that bet. Goldman Sachs has estimated that token consumption from AI agents alone will grow 24 times by 2030, a projection driven by the expectation that autonomous AI systems will increasingly execute multi-step workflows — booking travel, writing and deploying code, managing customer service interactions — that generate vastly more API calls than a human sitting at a chat interface.
How 1Password’s new dashboard tracks every token across Anthropic, Cursor, and OpenAI
The new capability extends 1Password SaaS Manager‘s existing foundation of application discovery, license management, and spend analytics. It is not a standalone product. Existing SaaS Manager customers can activate it by connecting their supported AI vendor API keys, at which point consumption data flows into a dedicated AI Consumption Management dashboard. Henry confirmed that there is no separate product or add-on fee: “AI Spend and Consumption Management is available to all 1Password SaaS Manager customers.”
The system provides four core functions. First, it aggregates token usage and spend across Anthropic, Cursor, and OpenAI into a single, normalized view — eliminating the need to toggle between three separate vendor dashboards with three different reporting formats. Second, it enables budget controls: organizations can set vendor-level spend limits, configure percentage-based thresholds, and receive automated alerts when prepaid balances approach depletion. Third, it disaggregates consumption by team, user, vendor, and model, allowing finance and IT to understand not just how much is being spent, but where and by whom. Fourth, it situates AI spend within the broader SaaS portfolio, helping organizations see how token costs relate to their total software investment.
Notably, the system captures consumption regardless of whether a human or an AI agent generated it. “Token consumption is captured at the API level regardless of whether a human or an agent is generating it,” Henry explained. “Organizations get the total consumption picture, including the spikes that agent loops can create, which can be some of the hardest usage to catch before it becomes a problem.”
Advertisement
That agent-level visibility matters because autonomous AI systems can generate runaway costs in ways that human users typically cannot. An agentic coding assistant stuck in a retry loop, for example, can consume thousands of dollars in tokens in minutes — with no human in the loop to notice. For now, the product alerts but does not enforce. When asked whether 1Password will eventually give organizations the ability to automatically cut off spending when a threshold is crossed, Henry said the company is “actively evaluating” automatic enforcement but emphasized that visibility must come first: “You can’t enforce what you can’t see.”
The choice of launch partners reveals where enterprise AI budgets are under the most pressure
The decision to start with Anthropic, Cursor, and OpenAI — rather than casting a wider net — reflects where enterprise AI adoption and budget strain are most concentrated right now. Henry said the choice was driven entirely by customer demand. “Anthropic, Cursor, and OpenAI are where we’re seeing the highest adoption, and where token consumption can move fast and get ahead of the teams responsible for managing it,” he said. The company plans to add additional vendors based on customer demand, API availability, and budget impact, though it has not committed to a specific timeline or vendor list.
The inclusion of Cursor alongside the two major foundation model providers is telling. Cursor, an AI-powered code editor that has rapidly gained traction among developers, represents a category of AI tool where consumption is particularly difficult to forecast. Unlike a chatbot interface where a user consciously types a prompt, Cursor integrates AI suggestions directly into the development workflow, generating token consumption continuously as developers write code. That ambient, always-on consumption pattern makes it especially prone to budget overruns.
Henry also addressed who inside an organization should actually own this problem — and acknowledged that the honest answer right now is no one. “When spend is fragmented across vendor dashboards and finance teams are reconciling it monthly, you’re always behind,” he said. “AI spend can’t be treated as a finance-only or IT-only problem.” He noted that the pricing differences between models have become significant enough that the choice of which AI model a team uses is now a meaningful financial decision, one that is pulling CFOs into conversations with IT, product, and engineering leaders “in ways they never had to before.”
Advertisement
Steve May, director of IT at ServiceTrade, a 1Password customer that has been using the capability, said it addressed a concrete planning gap. “Forecasting tools for AI consumption and spend was one of our biggest gaps in planning because we didn’t have a reliable way to track it,” May said. He added that the visibility has “prevented overages that would have cost far more to fix after the fact.”
Where 1Password fits in the fast-consolidating SaaS management market
1Password is not the only company racing to solve the AI cost management problem, but the competitive landscape is still fragmented and the category is far from mature.
Zylo, a SaaS management platform that Gartner has also recognized as a leader in the space, published its 2026 SaaS Management Index in January showing that AI-native application spend surged 393% year over year in organizations with more than 10,000 employees and 108% overall. Zylo’s data also revealed that ChatGPT has become the most expensed application in enterprise environments, highlighting how AI tools are entering organizations through employee credit cards and expense reports — outside formal procurement and governance workflows. Zylo has added its own token-level cost tracking for AI vendors including Anthropic, OpenAI, Cursor, and Perplexity.
Meanwhile, according to a comparison published by Coommit in May, Vendr — which focuses more on SaaS negotiation than discovery — tracks AI tools at the contract level but does not yet offer consumption-level visibility. And the FinOps Foundation reported in its 2026 State of FinOps survey that 98% of organizations now actively manage AI costs, up from just 31% in 2024. The broader SaaS management market is also consolidating rapidly. In May, Deel acquired Sastrify, a German SaaS management vendor, and began folding it into its HR platform — a signal that SaaS management capabilities are increasingly being absorbed into adjacent enterprise platforms rather than remaining standalone products.
Advertisement
1Password’s approach differs from pure-play SaaS management competitors in one important respect: it is building AI cost management on top of an identity security platform, not a FinOps or procurement tool. The company’s SaaS Manager product grew out of its 2025 acquisition of Trelica, a UK-based SaaS access management startup whose technology enabled the discovery of unsanctioned applications — so-called shadow IT. As BetaKit reported at the time of that deal, 1Password co-CEO Jeff Shiner described Trelica as “a pioneer in modern SaaS access management” and said the acquisition would accelerate 1Password’s Extended Access Management product roadmap by more than a year. CRN noted that Trelica brought more than 300 SaaS integrations to the platform. That identity-first lineage gives 1Password a natural advantage in connecting spend data to specific users and teams — a linkage that matters when the question shifts from “how much are we spending on AI?” to “who is spending it, and is it delivering value?”
From password manager to platform company: 1Password’s $6.8 billion bet on enterprise identity
The launch raises a question that Henry addressed head-on: whether a company that started as a consumer password manager can credibly compete in enterprise AI cost management.
“It doesn’t feel like a stretch to us. It feels like a natural progression,” he said. “For more than 20 years, 1Password has evolved alongside how our customers work. We started by protecting passwords. Then we helped organizations manage secrets, control access, and get visibility into the applications their teams rely on.”
The company’s evolution has been rapid. 1Password raised a $620 million Series C in January 2022 led by ICONIQ Growth, reaching a $6.8 billion valuation — at the time, the largest funding round ever raised by a Canadian company, according to Crunchbase. The round also attracted celebrity investors including Ryan Reynolds, Scarlett Johansson, and Robert Downey Jr. As of early 2025, BetaKit reported that 1Password had surpassed $250 million in annual recurring revenue, with B2B sales accounting for nearly three-quarters of total revenue and the company claiming to be cash-flow positive.
Advertisement
In May 2024, 1Password launched Extended Access Management, a platform designed to secure sign-ins across both managed and unmanaged applications and devices. That same year, it acquired Kolide for device trust and, in early 2025, Trelica for SaaS discovery. In June 2026, Gartner named 1Password a Leader in its Magic Quadrant for SaaS Management Platforms. According to 1Password’s own blog post on the recognition, its SaaS Manager now supports over 400 integrations and provides visibility into a library of more than 40,000 pre-populated application profiles. Each step has moved the company further from its consumer roots and deeper into enterprise infrastructure. The AI Spend and Consumption Management launch extends that trajectory into financial operations territory — a domain where 1Password will compete not only with SaaS management vendors but potentially with dedicated FinOps platforms and the AI vendors’ own billing dashboards.
Why high AI token consumption doesn’t always mean wasted money
Perhaps the most revealing part of Henry’s commentary concerns what organizations should actually do with the consumption data once they have it. He pushed back forcefully against the assumption that high token consumption automatically signals waste.
“A team burning through tokens may be building something genuinely valuable,” he said. “A lower-usage project might not be moving the business forward at all. What matters is whether that consumption is producing enough business value to justify the spend.”
Henry drew a distinction between personal productivity — “having a bot summarize your meeting or draft a quick email” — and genuine business outcomes. “What organizations need to see is where consumption is actually driving revenue, efficiency, or something that moves the needle.”
Advertisement
That framing positions AI Spend and Consumption Management not just as a cost-cutting tool but as a decision-support system for AI investment allocation. If a CFO can see that one engineering team’s heavy Claude usage is powering a product feature that drives revenue, while another team’s OpenAI spend is funding low-value internal automation, the organization can reallocate budget accordingly rather than imposing across-the-board cuts.
“When costs rise faster than expected, the instinct is to cut,” Henry said. “But most organizations can’t yet tell which teams, models, or tools are responsible for the increase, so they end up cutting across the board rather than directing investment toward the AI projects that are actually delivering business value. Blunt cuts on a technology you’re counting on for competitive advantage is not a management strategy, it’s a missed opportunity.”
The next enterprise budget crisis is already here — and it’s priced per token
The product’s current scope — three vendor integrations, alerting but not enforcement — is clearly a starting point. Henry signaled that automatic spend limits are on the roadmap and that additional vendor integrations will follow based on customer demand.
But the broader trajectory he described suggests 1Password sees this launch as a wedge into a much larger opportunity. “As traditional SaaS products add AI capabilities, their pricing models are going to follow,” he said. “Organizations that build visibility and management discipline around consumption now are going to be in a much better position when that happens across the rest of their software portfolio.”
Advertisement
If Henry is right, the chaos currently confined to AI token budgets is not a temporary growing pain but a preview of how all enterprise software will eventually be priced. A decade ago, companies scrambled to understand their cloud bills. Today, they are scrambling to understand their AI bills. The question is whether the organizations building the dashboards this time around can get ahead of the curve — or whether, as Henry warned, they will end up where so many companies ended up with cloud, realizing too late how much they were overpaying, and for how long.
AI Spend and Consumption Management is available now in public preview for 1Password SaaS Manager customers. Broad availability is planned for fall 2026.
Bit.Bio’s Dr Emma Pepperell discusses her career in the biotechnology sector, the advancements occurring around her and the importance of bringing a touch of humanity to an often clinical industry.
Interested in veterinary medicine and science from a young age, the now Dr Emma Pepperell focused on the subjects in her early education, before committing to a BSc in Pharmacology at Newcastle University, where she graduated with first class honours.
It was during her undergraduate degree that she undertook a placement at pharmaceutical GlaxoSmithKline, where for a year she spent her time working with organ models of the brain and gut, to better understand how one might impact the other.
Pepperell said, “Even then, 20 odd years ago, there was the idea that what’s going on in the brain could affect things like Crohn’s disease and conditions like this that we know affect a lot of people.
Advertisement
“So that sort of sparked my interest in drug development and the whole sort of pharmaceutical industry and the application of science and research products for health.”
Inspired to continue her education in this field, Pepperell enrolled in the University of Oxford, where she soon graduated with a DPhil in Clinical Laboratory Sciences. As so often happens, her Doctorate focused on an area of learning that would eventually form the basis of her career, which in this case was the application of stem cells to therapies.
She said, “I was looking at cells from umbilical cord blood to see if they could repopulate the bone marrow for childhood leukemia, so it was very translational. And then I went into the commercial side of life sciences.
“The opportunity to join Bit.Bio came up a couple of years ago and it seemed like a natural, sort of complete circle to go back to the stem cell field and really think about how could I apply everything I’d learned about life science research tools and drug development and to think about how we can really bring these human cell products to have an impact, help develop drugs and understand disease better.”
Advertisement
For others looking to move into the STEM space or even careers outside of it, she finds it crucial that no matter what decisions you make, that you show a genuine interest in what you are doing. If you have a natural curiosity about something, she finds you will always be far more motivated and invested in developing yourself to fit the dream.
“I would say for young students starting their career, try and try lots of work experience and find what works for you. I went to a veterinary practice, I went to a patent attorneys and learned about patenting screwdrivers. I tried all sorts of different things to see, okay, what is it that’s actually going to really interest me, so I can find something that’s going to be highly motivating.”
Keeping healthcare humane
Having contributed to the work at Bit.Bio for two years, Pepperell was recently named the organisation’s CEO.
She said, “Fundamentally, what we do is we make human cells. So you can take an adult skin cell or blood cell and reprogramme it back to what’s called a pluripotent state and that means it can become any cell in the body.
Advertisement
“So we can then put in codes that essentially direct the cell to become a neuron, a brain cell, a liver cell, a heart cell, any kind of cell you like and then you can use these cells to really start to understand how disease works.
“You can use them to identify new drug targets. You can use them to study what happens if a new drug that’s being developed is applied to them to see if that drug is toxic or what the metabolic profile is like. We make cells that essentially replicate human biology in quite a controlled way to really enable the study of health and disease and drug interactions.”
An element of her industry that greatly appeals to her is largely in how it manages the challenges that arise. Pepperell explained how, in drug development, animal testing is an unfortunate, albeit often necessary and even crucial component of the drug development process.
As more organisations and institutions look to methodologies that limit the use of animal models in creating results for human biology, Pepperell said, “You know we use animal models because it represents a system, but nobody really wants to use them where they can avoid it.
Advertisement
“So, there is a big challenge in the industry at the moment for building non-animal models that people can trust will deliver safe drugs to the market, whilst acknowledging that all of the knowledge and evidence that has been built up for almost the entirety of the pharmaceutical industry’s history has been in animal models.”
She noted her pride at being part of an organisation that works to support scientists in the adoption and development of new approaches, required by the FDA and other bodies, that may ultimately phase out the use of animals wherever possible in drug development.
Keeping such a clinical industry humane, is for Pepperell, particularly important, as her mission and that of Bit.Bio’s is to leave society better off than they found it. Knowledge sharing is one such way she believes STEM professionals can contribute to a fairer and more thoughtful world.
She said, “You are only as successful as the teams who are around you because no one can achieve anything on their own.”
Advertisement
An ideal example, she finds, is Bit.Bio’s recently organised Human Cell Forum, where the company brought together 200 scientists all working on human models designed for better drug development. She found that the community feel it created, particularly amongst professionals that are often under immense pressure to secure funding and resources, in some ways alleviated silos and lead to a positive outcome in the research space.
She said, “What we start to see in these events that we’ve put together is that scientists are really coming together to share how they’re solving problems. One example is ALS, which is a neurodegenerative disease.
“There were three or four companies who are all trying to develop a similar model and they all brought their learnings to this event and we’re very openly sharing with each other how they’ve optimised the model and that’s quite rare to see.”
Ultimately, she finds it often boils down to who you have in your corner, having faith in your own skills and striving each day to do better than you did the day before, not just for yourself, but for the colleagues, patients and test animals impacted by your work.
Advertisement
Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.
Do you remember AIM? It may suprize you to hear that AOL’s instant messanger was actually supported all the way up to 2017– two years after Discord launched. Unlike Discord, AIM is a protocol, not a platform. Everything on your favourite Discord server is at the mercy of the corporate masters of said server; you can’t just spool up your own. Not so for AIM, as [Veronica] explains, both on her blog and in a YouTube video that we’ve embedded below.
The key is the fact that the AIM protocol isn’t locked into AOL’s now-defunct servers; it was reverse engineered in its prime for open-source messengers like Pidgin. You can host your own server, too, using the OpenOscarServer by [mk6i]. Even better, it’s not just AIM, but ICQ! In the sort of irony you only get in real life, the OpenOscar community does all its support on a Discord server. But then, they couldn’t hardly do it over AIM or ICQ these days.
For those of you who were too old or too young to get sucked into the 90s instant messenger craze, these protocols don’t just create chat rooms, that would be the even older Internet Relay Chat protocol, but usually worked more like SMS text messages. You have a contact list, and you send messages to your contacts via a server that acts as a hub. Once upon a time, that server was AOL’s, but now thanks to the OpenOscar project, it can be anybody’s computer. Of course, like texting, you can rope all of your contacts into one big group chat, and the protocol does support images and VOIP. (Which is starting to sound a lot like Discord.)
If you’re tired of your friend-group being at the mercy of American tech companies, [Veronica]’s blog post serves as a good guide to get you started running OpenOscarServer on a Linux system; she used a virtual private server but figures a Raspberry Pi ought to have enough grunt if you don’t have a huge number of people signed up.
Advertisement
For completeness, we should mention that while AOL pulled the plug on AIM nearly a decade back, ICQ, the other protocol supported by OpenOscarServer, lasted straight through until 2024.
Thanks to Keith Olson for the tip! Our tipsline is based on decentralized “electronic mail” technology that anyone can access.
Even though its price has climbed steadily in the last decade, the Ford Mustang GT is still one of the more exciting performance cars on the market. There are lots of reasons buyers might be drawn to it, and its 5.0-liter V8 engine is surely one of the big ones. A naturally aspirated V8 once represented the backbone of the American performance car scene, and with Chevy axing the Camaro and Dodge’s V8 models all currently on hiatus, the Mustang has become the only (relatively) affordable game in town.
However, part of the reason V8s are otherwise so few and far between is that manufacturers have increasingly shifted to smaller-displacement engines, including turbocharged V6s. And in some cases, these V6s can actually outgun the 480-hp V8 Mustang GT in straight-line performance.
Advertisement
The Mustang’s acceleration will vary depending on which transmission it has, but the average figures for the GT show a 0-60 time in the low to mid four-second range and a quarter-mile run in the low to mid 12-second range. Those are stout numbers, and for many buyers, the Mustang GT’s V8 soundtrack is non-negotiable — but if you compare the Mustang to modern V6-powered performance cars, you’ll find several that can edge it out in straight-line acceleration.
Advertisement
Cadillac CT4-V Blackwing
With General Motors discontinuing the Chevrolet Camaro, those looking for a front-engine, rear-drive GM performance car now have to move up to Cadillac’s Blackwing-branded sport sedans. When it comes to V6-powered performance sedans, the Cadillac CT4-V Blackwing is one of the hottest around, with its twin-turbocharged 3.6-liter engine rated at 472 hp and 445 lb-ft of torque.
The CT4 has slightly less horsepower than the Mustang GT, but at the track it can hit 60 mph in four seconds flat and run the quarter mile in 12.4 seconds, with the automatic being slightly quicker than the manual version. While the acceleration battle between the two cars is close, the Mustang GT gets the win in the value department with its sub $50,000 starting MSRP coming in substantially cheaper than the CT4-V Blackwing’s mid $60,000s base price. This isn’t surprising given the Blackwing’s more luxurious persona and branding.
The Cadillac may not have the V8 engine that so many associate with American performance cars, but the numbers show the CT4-V Blackwing is more than capable of running with larger-displacement rivals. Better yet, our review of the manual transmission-equipped CT4-V Blackwing also showed the car to have an engagement and fun factor that goes beyond its performance figures.
Advertisement
Nissan Z Nismo
As a two-door rear-drive sports car from a mainstream brand with a starting price in the mid-$40,000s, the Nissan Z is actually one of the Mustang’s most direct competitors in this group. And unlike other cars, which have downsized their engines in the modern era, a V6 engine has been a key part of the Nissan Z formula going all the way back to the mid-1980s. Today, all versions of the Z are powered by 3.0-liter twin-turbocharged V6 engine, but it’s the high-performance Nismo variant that makes the most of that V6 powerplant.
In its Nismo trim, the Z makes 420 hp and 384 lb-ft of torque — and when mated to an automatic transmission, testing has shown that combo is good for a 0-60 run of 3.9 seconds and a quarter-mile time of 12.4 seconds. These are impressive figures given the Z’s horsepower figures are actually fairly modest by mid-2020s performance car standards.
Advertisement
While we had some mixed feelings about its price, our review of the Nissan Z Nismo, showed that this machine has a lot to offer for fans of modern Japanese sports cars. If the Nismo Z’s price seems too high, the cheaper, non-Nismo version isn’t far off performance-wise, with low four-second 0-60 times and high 12-second quarter-mile ETs that put it pretty close to the Mustang GT.
Advertisement
Audi S5 and RS5
Comparing the Mustang GT to the Audi S5 shows just how varied the modern performance car can be. The cars are of a comparable weight and size, but that’s about where their similarities end. The Mustang GT has a naturally aspirated V8 and rear-wheel drive, while the S5 has a twin-turbocharged V6 and all-wheel drive — and a price that starts in the mid $60,000s.
Rated at 362 hp, the S5’s 3.0-liter twin-turbocharged V6 engine is down by over 100 hp compared to the Mustang, but it claws back an acceleration advantage at the track thanks to that aforementioned all-wheel-drive system. Testing has shown the S5 to hit 60 mph in just 3.9 seconds, with the quarter-mile coming in 12.5 seconds.
Our review of the S5 Sportback showed the car to be a competent and quick luxury machine, but if that’s not enough for you, there’s always the new Audi RS5. The RS5 also has a twin-turbo V6 engine, but adds a plug-in hybrid electric boost for a total of 630 hp. This makes the car good for 0-60 times in the low three-second range and a quarter-mile time in the mid 11s. Just know that you can purchase two new Mustang GTs for the Audi’s $100,000-plus starting price.
Advertisement
Nissan GT-R R35
You could say the R35 Nissan GT-R is an unfair addition to this list. For starters, you can’t actually buy a new one anymore, as Nissan ended production of the GT-R in 2025. And, going back to its debut in the late 2000s, the GT-R always played in a completely different segment than the Mustang GT, with a much higher price, and its sights set on high-end European supercars.
However, as an enduring symbol of the V6 engine’s performance potential, the GT-R is more than deserving. Earlier iterations of the GT-R used the legendary RB26 inline-six, but the R35’s switch to a new 3.8-liter twin-turbocharged engine showed just how capable a V6 could be. The R35 GT-R was in production for a long time, with a lot of updates along the way — and the Nismo variants of the car were rated at an impressive 600 hp and 481 lb-ft of torque.
Advertisement
At the track, the GT-R was capable of hitting 60 mph in 2.9 seconds and running the quarter-mile in the low 11s. Price aside, those are incredible numbers for a V6-powered car that was originally developed back in the 2000s. By the time Nissan pulled the plug on the R35, the GT-R was showing its age in many ways, but its raw performance figures still hold up against today’s best, no matter how many cylinders they may be packing.
Advertisement
Porsche Macan GTS
Looking back through the Porsche brand’s long history of building performance cars, the V6 probably isn’t the first engine type that comes to mind, but it’s what powers — or powered – the higher-end versions of the brand’s popular Macan crossover SUV in recent years. With V6 power, the Macan is one of the more potent and practical performance machines on the market.
The high-performance Macan GTS variant gets its power from a 2.9-liter twin-turbo V6 that makes 434 hp and 405 lb-ft of torque, sending that power to all four wheels. In performance testing, the gasoline Macan GTS delivers, hitting 60 mph in 3.5 seconds and running the quarter-mile in just 12.1 seconds. On top of that, it also has handling that blurs the lines between crossover SUV and serious sports car.
In the real world, are there any buyers seriously cross-shopping any trim of the Porsche Macan against any trim of the Ford Mustang? Most likely not, but the fact that this V6-powered crossover SUV can show the V8 Mustang its taillights at the drag strip shows just how much the performance car has evolved in the modern era. Speaking of evolution, the gasoline-powered Macan is now on its way out, to be replaced by the new all-electric Porsche Macan, which has some big shoes to fill.
UN Secretary General calls for a global ban on autonomous “killer robots”
Guterres argues that delegating life-or-death decisions to machines is “morally repugnant”
Governments should take a stance now – not wait for something catastrophic to happen
UN Secretary General António Guterres has called for lethal autonomous weapons, which he describes as ‘killer robots’ to be prohibited under international law following recent discussions at the first Global Dialogue on Artificial Intelligence Governance in Geneva.
Guterres’ demand to ban these weapons focuses on those capable of identifying, selecting and attacking targets without human oversight, which leaves artificial intelligence and other computer systems in charge of a life-or-death decision.
He ultimately argued that certain decisions must remain exclusively human, and the decision to take a life is well into the boundary of requiring human oversight. Transferring the decision-making to killer robots would be “morally repugnant” and “politically unacceptable,” he argued.
Latest Videos From
AI requires global regulation as military AI poses major threats
Key to the Secretary General’s argument is that he urges governments to take action and ban such robots now, rather than waiting for an autonomous weapon to cause a major incident before rethinking their strategies.
Advertisement
“Let us not wait for atrocity to act,” Guterres said. “Some decisions must remain forever human – none more than taking a human life.”
The issue is becoming more urgent now that AI models and advanced chips are already being used within military intelligence, targeting and other battlefield systems.
More broadly, Guterres’ thoughts align with those of Anthropic, which recently had a dispute with the Pentagon after seeking guarantees that its models would not be used for autonomous weapons or surveillance.
Advertisement
Sign up to the TechRadar Pro newsletter to get all the top news, opinion, features and guidance your business needs to succeed!
While the Pentagon had rejected those limitations, arguing that it should be able to use Anthropic’s models for any lawful purpose, the case highlights how private companies are becoming increasingly intertwined with digital warfare.
Reporting by the Wall Street Journal cited a similar view by Pope Leo XIV, who warns that AI-controlled weapons could promote an “anti-human” view of warfare. He warned that the autonomy could reduce some dangers and distance political leaders from the human consequences of conflict.
Advertisement
There’s a need to balance the pros and cons of AI
However, artificial intelligence does promise several benefits to modern warfare, particularly in its ability to process huge amounts of information extremely quickly. With modern compute, militaries can respond to threats at lightning speed, improve their accuracy and precision, reduce soldier risk and potentially reduce civilian casualties, too.
Critics also question whether human oversight of AI systems is at all meaningful if the person in charge only has seconds to act on AI-generated information in the first place.
It’s also yet to be determined which party or group of parties should be held accountable for any incidents or mishaps – human operators, commanders, hardware manufacturers and software developers are just some of the parties up for judgment.
Advertisement
“We may be the last generation able to set the terms on which humanity and machines coexist,” Guterres warned separately in an X post, warning that AI must be governed, trusted and fair.
“It sounds like science fiction, but it’s a real possibility, and it could change the world in ways that we don’t understand yet, and it could change the power dynamics of our planet in ways that require our attention,” Independent International Scientific Panel on AI Co-Chair Yoshua Bengio added.
In X Square Robot’s approach, demonstrations can be replayed and executed on a physical robot through the AI model, allowing the same data to be reused across different platforms.X Square Robot
X Square Robot’s approach is that, in a real home, failure recovery matters more than raw success, because the home does not reset itself and it demands careful personalization, with safety and trust as first principles. X Square Robot
You must be logged in to post a comment Login