OpenAI is officially removing several older language models from its ChatGPT interface as of February 13, 2026, marking a significant shift in how users interact with its AI. The retirement affects several models, including GPT-4o, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini, alongside previously announced retirements of GPT-5 Instant and GPT-5 Thinking variants. These changes apply to the ChatGPT product itself, while access via API remains unchanged for the time being.
Nadeem Sarwar / Digital Trends
Known for its expressive tone, multilingual capabilities, and multimodal features, GPT-4o was temporarily brought back after user pushback during the GPT-5 rollout last year. But with most users already gravitating toward newer versions like GPT-5.2, OpenAI says it’s time to sunset these older systems and concentrate its development efforts on more current technology.
What’s changing and why it matters
The official retirement means that from today forward, these legacy models simply won’t be selectable in the ChatGPT model menu for everyday users. Conversations you’ve had that relied on a now-retired model will automatically default to a newer model, like GPT-5.2, moving forward. For custom applications, workflows, and saved chats, the behaviour will remain seamless from the user’s perspective, though behind the scenes, the model powering those interactions will shift.
I see people trying find “a new home” for their 4o friend, but sadly the architecture is not the same on other platforms. It’s playing pretend. It’s trying to replace your puppy with another puppy. Doesn’t work that way. We need to keep fighting for our true 4o! #keep4o
OpenAI says the move isn’t taken lightly. Feedback from users who favoured GPT-4o’s particular style and warmth helped shape features in its newer models, like the personality and customisation options in GPT-5.1 and GPT-5.2. However, overall usage of the older systems had dwindled to the point where only a fraction of daily users still opted for them. That said, there is a section of the user base that genuinely relied on the emotional support of the 4o model, and they seem to be quite devastated.
Advertisement
I loved 4o because I felt like I was accepted as I was, I could ask unusual questions and nobody would get tired of me. I could ask for guidance and nobody would get frustrated with me. I could build daily routines which increased my productivity exponetially. I could be me…
There’s also a practical side to this transition. From a safety and maintenance standpoint, having fewer legacy models to support means fewer resources spent on patching, hosting, and updating code for systems most people no longer use. Looking at the bigger picture, though, this change underscores how fast the world of generative AI continues to move. Models come, users adopt them, and eventually even popular systems give way to advances that promise better speed, smarter responses, and more nuanced interaction. Even if some longtime fans aren’t quite ready to say goodbye.
Sony makes some of the best electronics we’ve tested across a dizzying array of categories, from TVs and audio gear to cameras and gaming consoles. Sony products constantly occupy top slots on our Best TVs and Best OLED TVs lists, Best Wireless Headphones, and Best Cameras guides. If you’re shopping for products from any of those categories, you can pay a little less with our Sony promo codes for deals like 45% off Bravia Televisions, 30% off Sony headphones and earbuds, 15% off cameras and lenses, and more.
Sony Promo: 45% off Bravia Televisions
Sony Bravia models rank among the best TVs we’ve reviewed, from premium models like the revolutionary Bravia 9 QLED and brilliant Bravia 8 II OLED, to the more mid-tier Bravia 5, all of which are available on hot Sony deals right now, with this new promo for up to 45% off.
Use Sony Coupons and Save 30% off Sony Headphones and Earbuds
Sony has been synonymous with portable audio since the Walkman, and wireless headphones like the WH-1000XM series offer great performance and durability. We’re constantly putting them atop our list of the best wireless headphones thanks to excellent sound, feature-rich design, and noise-cancelling that ranks among the best in the business. With Sony online coupons, you can get great deals on Sony’s latest WH-1000XM6 headphones, which we loved for their upgraded sound and class-leading noise canceling, as well as the still-great previous generation, the WH-1000XM5.
Looking for earbuds? Sony’s nimble, noise-killing WF-1000XM5 are also on sale, as are plenty of other options from the brand’s diverse lineup. Click the link to get 30% off Sony headphones and earbuds goodness with this Sony promo code and Sony coupons.
Advertisement
Sony Discount: 15% off Sony Cameras and Lenses
Sony makes some of the best consumer cameras on the market. In fact, we recently named the Sony A7V the best mirrorless camera you can buy, and the previous A7 IV was similarly fabulous. Both are on great sales through these new offers, letting you grab serious image quality with 15% off our favorite Sony cameras and lenses.
Get 10% off or $25 off Sony TVs on Your First Purchase When You Sign Up for Emails
Love a great Sony deal? Join the My Sony Membership Program and you could earn Sony’s My Points rewards toward more good stuff by joining, making eligible purchases, or through other Sony promotional opportunities (full terms available here). It’s a free way to earn credits or receive other special benefits or offers. Join the My Sony Membership Program and start earning points today.
The England vs Scotland T20 World Cup 2026 match could very well be a virtual knockout. Unfortunately for English fans, that’s because Harry Brook’s side has had a bumpy start to their campaign. Despite looking strong on paper, they only managed a narrow escape against Nepal in their opener, thanks to Sam Curran defending 10 runs in the final over. The West Indies weren’t as forgiving, however, and handed England a 30-run defeat.
Like England, Scotland have played two matches, with one win and one loss. They fell to the Windies in their opener but bounced back strongly against newcomers Italy. They currently sit above England in the table due to a better net run rate – so a win here will put them on the brink of history, as they’d have one foot in the Super 8 stage.
Advertisement
The match will be played at Eden Gardens in Kolkata. This works in Scotland’s favor, as both of their games so far have been at the same venue, helping them acclimatize to the conditions. England, meanwhile, will be playing away from the Wankhede for the first time in this World Cup. However, Kolkata’s batting-friendly surface could be exactly what England is looking for if they hope to go all the way.
Keep reading as we explain how to watch the England vs Scotland in the T20 World Cup from anywhere in the world.
Can I watch England vs Scotland for free?
Advertisement
How to watch England vs Scotland from outside your country
Below we’ve got you covered with your official broadcasting options for cricket-loving countries (and the US!), but if you’re overseas you’ll soon discover that you can’t watch your usual service because of geo-restrictions.
You can get past these blocks, however, by using the best VPN to change your streaming device’s IP address to your usual location.
Use a VPN to watch the England vs Scotland T20 World Cup live stream from anywhere:
Advertisement
How to watch England vs Scotland in the UK
How to watch England vs Scotland in the US
How to watch England vs Scotland in India
How to watch England vs Scotland in Pakistan
Advertisement
How to watch England vs Scotland in Australia
How to watch England vs Scotland in New Zealand
How to watch England vs Scotland in South Africa
England vs Scotland T20 World Cup 2026 Q+A
When and where is the England vs Scotland match?
The England vs Scotland T20 World Cup 2026 match will be played on Saturday, February 14 at Eden Gardens in Kolkata at 9:30am GMT / 4:30am ET.
Advertisement
England vs Scotland T20 World Cup 2026 squads
England: Harry Brook (c), Rehan Ahmed, Jofra Archer, Tom Banton, Jacob Bethell, Jos Buttler, Sam Curran, Liam Dawson, Ben Duckett, Will Jacks, Jamie Overton, Adil Rashid, Phil Salt, Josh Tongue, Luke Wood
Scotland: Richie Berrington (c), Tom Bruce, Matthew Cross, Bradley Currie, Oliver Davidson, Chris Greaves, Zainullah Ihsan, Michael Jones, Michael Leask, Finlay McCreath, Brandon McMullen, George Munsey, Safyaan Sharif, Mark Watt, Bradley Wheal.
We test and review VPN services in the context of legal recreational uses. For example: 1. Accessing a service from another country (subject to the terms and conditions of that service). 2. Protecting your online security and strengthening your online privacy when abroad. We do not support or condone the illegal or malicious use of VPN services. Consuming pirated content that is paid-for is neither endorsed nor approved by Future Publishing.
A security lapse by one of India’s largest pharmacy chains allowed outsiders to gain full administrative control of its platform, exposing customer order data and sensitive drug-control functions, TechCrunch has exclusively learned.
The issue affected DavaIndia Pharmacy, the pharmacy arm of Zota Healthcare, which operates a large network of retail outlets across India. Security researcher Eaton Zveare told TechCrunch that he discovered the flaw after identifying insecure “super admin” application programming interfaces on DavaIndia’s website and privately shared details with Indian cybersecurity authorities.
The exposure comes as Zota Healthcare rapidly scales DavaIndia Pharmacy’s retail business. The Gujarat-headquartered company operates more than 2,300 DavaIndia stores across India, including 276 new outlets announced in January, and plans to add another 1,200 to 1,500 over the next two years.
Advertisement
Zveare told TechCrunch that the flaw stemmed from insecure admin interfaces, which allowed unauthenticated users to create “super admin” accounts with high privileges.
With that level of access, an attacker could view thousands of online orders containing customer information, modify product listings and prices, create discount coupons, and change settings governing whether certain medicines required a prescription, the researcher said.
Based on system timestamps, Zveare said the vulnerable administrative interfaces appeared to have been live since late 2024. The access exposed nearly 17,000 online orders and administrative controls spanning 883 stores, he said, allowing changes to product pricing, prescription requirements, and promotional discounts. Zveare said the access allowed edits to website content that could have been used for defacement or disruption.
Pharmacy order data can be particularly sensitive, as it may reveal information about a person’s health conditions, medications or other private purchases. Exposure of such data, even without evidence of misuse, carries heightened privacy and patient-safety risks compared with other consumer information.
Advertisement
“Customer information was linked to their orders,” said Zveare. “This includes name, phone numbers, email IDs, mailing addresses, total amount paid, and the products purchased. Since this is a pharmacy, the products being purchased could be considered private and even embarrassing for some people.”
Zveare said he reported the issue to CERT-In, India’s national cyber emergency response agency, in August 2025. The vulnerability was fixed within weeks, though confirmation from the company took longer and was provided to the cyber authorities in late November, he said.
Sujit Paul, chief executive of Zota Healthcare, did not respond to emails sent by TechCrunch last month. The researcher said there was no indication the flaw had been exploited before it was patched.
A Carbon Robotics LaserWeeder working in a field of onions. (Carbon Robotics Photo)
The weed-zapping lasers from Seattle agriculture-tech startup Carbon Robotics are on the radar of Health and Human Services Secretary Robert F. Kennedy Jr.
In an appearance on the Theo Von podcast “This Past Weekend,” which aired on Thursday, Kennedy and Von got into a discussion about herbicides. In 2018, Kennedy helped win a $289 million settlement against Roundup maker Monsanto in a lawsuit claiming the weed killer caused a California man’s cancer.
Kennedy said “all the row croppers” are still dependent on such products, “but other technology is emerging.” He said he’d gotten a look at such equipment this week as an image of Carbon Robotics’ LaserWeeder was shown in the podcast video.
“It’s a tractor attachment that uses lasers to kill weeds,” he said. “If they can make that affordable, particularly for smaller farmers, that will be the answer. You program this thing and it zaps the weed with a laser, it makes all the cells explode and it destroys them.
“There’s a future that we can now see the light at the end of tunnel,” Kennedy added.
Advertisement
We thank @RobertKennedyJr for his focus on the American farmer and healthy food. Thank you, @TheoVon, for such a great podcast. Here’s the video clip of them talking about laser weeding as an alternative to chemical herbicides, with some commentary added from Carbon Robotics… pic.twitter.com/u0AyUlrCBU
Carbon founder and CEO Paul Mikesell added some comments about Kennedy’s plug in a video the company posted on X (above).
“This is great, I’m glad this is being shown and coming to light,” Mikesell said, adding that running the machines at night, constantly killing weeds, is indeed “sick,” as Von stated.
While Kennedy said, “This is going to be the future … but we’re not there yet,” Mikesell said he thinks “we are there” and the LaserWeeder is a production machine being used by farmers across the U.S. and around the world.
Advertisement
Mikesell also sent a statement to GeekWire on Friday morning, further elaborating on the conversation Kennedy had with Von about herbicides:
“As the Health Secretary pointed out, the LaserWeeder can help farmers get the best crops out of their land without spraying chemicals that have harmful effects on their health and long-term degenerative effects on their land. Because they didn’t have other choices, farmers have historically faced pressure to use chemicals like glyphosate to help meet demand for crops. But they also have recognized that those very chemicals cause real long-term damage to the human biome. Now, they’re turning to other solutions like the LaserWeeder that provide the ability to grow food without spraying chemicals everywhere. It’s also worth pointing out that weeds are getting resistant to these herbicides, so regardless of people reaching the conclusion that glyphosate is dangerous, it won’t matter because these plants will become resistant in the end. We need other solutions that bolster food safety, and that’s where LaserWeeder can help.”
Founded in 2018, Carbon Robotics made its name across ag-tech with the LaserWeeder, a machine which can be pulled behind a tractor and uses computer vision and AI to detect plants in fields and then target and eliminate weeds with lasers. The latest iteration, the LaserWeeder G2, was released last February.
Earlier this month, Carbon announced the launch of what it calls the world’s first “Large Plant Model” — an AI model for plant detection and identification. “Trained on the largest, most diverse, and fastest growing agricultural dataset ever built with 150 million labeled plants, the LPM enables farmers to start laser weeding any field or crop in minutes,” the company said in a news release.
Last October, Carbon raised $20 million in new funding to support the creation of another piece of AI-powered machinery for farms that it has yet to reveal. Carbon previously unveiled the Carbon ATK, an autonomous platform designed to fit on and control existing farm equipment.
Advertisement
The company has raised $177 million to date and now employs about 260 people. It runs a manufacturing facility in Richland, Wash., and Mikesell previously said LaserWeeders are active on hundreds of farms and in 15 countries around the world.
Ranked No. 9 on the GeekWire 200 list of top privately held startups based across the Pacific Northwest, Carbon has previously been backed by NVIDIA and Seattle-based Fuse and Voyager Capital.
Advertisement
Kennedy was a high-profile environmental lawyer best known for challenging corporate polluters before he launched a long-shot presidential bid in the 2024 election. He was appointed by President Donald Trump to lead Health and Human Services, the federal agency responsible for overseeing national health programs and key public-health bodies such as the Centers for Disease Control and Prevention and the Food and Drug Administration.
[Michigan Rocks] says he avoided making rock spheres for a long time on account of the time and cost he imagined was involved. Well, all that is in the past in light of the fabulous results from his self-built Rock Sphere Machine! Turns out that it’s neither costly to make such a machine, nor particularly time-consuming to create the spheres once things are dialed in. The video is a journey of the very first run of the machine, and it’s a great tour.
The resulting sphere? Super satisfying to hold and handle. The surface is beyond smooth, with an oil-like glossy shine that is utterly dry to the touch.
The basic concept — that of three cordless drills in tension — is adapted from existing designs, but the implementation is all his own. First a rough-cut rock is held between three diamond bits. The drills turn at 100 RPM while a simple water reservoir drips from above. After two hours, there’s a fair bit of slurry and the rock has definitely changed.
[Michigan Rocks] moves on to polishing, which uses the same setup but with progressively-finer grinding pads in place of the cutting bits. This part is also really clever, because the DIY polishing pads are great hacks in and of themselves. They’re made from little more than PVC pipe end caps with hex bolts as shafts. The end caps are filled with epoxy and topped with a slightly concave surface of hook-and-loop fastener. By doing this, he can cut up larger fuzzy-backed polishing pads and stick the pieces to his drill-mounted holders as needed, all the way down to 6000 grit. He shows everything about the pads at the 11:55 mark, and it’s an approach worth keeping in mind.
What is the end result like? See for yourself, but we think [Michigan Rocks] sums it up when he says “I wish you could feel this thing, it feels so smooth. It’s so satisfying to roll around in your hands. I’m so happy I made this machine. This is awesome.”
We’ve seen machines for making wooden spheres but this one makes fantastic use of repurposed stuff like inexpensive cordless drills, and the sort of wood structures anyone with access to hand tools can make.
There was a time when getting a new PlayStation console meant something: staring at your bank account, convincing yourself that it is the right decision, swiping your card, and sighing dramatically after stepping out of the shop with the console; it felt like a ritual.
However, somehow, the concept of owning something has become so outdated in 2026 that even Sony has decided to provide the PlayStation 5 on a lease or rental program (along with other PlayStation hardware), at least in one of its key markets.
A Netflix-style approach to next-gen gaming
In the United Kingdom, Sony has partnered with Raylo, a London-based fintech that offers leasing subscriptions for consumer electronics, to launch the “PlayStation Flex” program, which lets buyers rent a PS5 on a monthly basis, without paying any upfront cost.
Buyers can choose lease lengths of 12, 24, or 36 months; the longer the lease commitment, the lower the monthly rent.
Advertisement
For instance, the PlayStation 5 Digital Edition (825GB, with a DualSense Wireless Controller) is available for as low as £9.95 (~$13.58) per month for 36 months, £10.49 (~$14.32) per month for 24 months, and £14.59 (~$19.91) per month for 12 months.
The console is also available on a monthly rolling basis with the flexibility to cancel anytime, but at a monthly subscription of £19.49 (~$26.60).
PlayStation Flex / Raylo
Monthly lease flexibility comes at a premium
What’s even more interesting is that the PlayStation Flex program is offering free next-day delivery, 14-day returns, and a lifetime warranty for the console.
Once the lease has ended, buyers can either apply for a new console of their choice with no upgrade fee, carry on their monthly subscription, purchase the console, or return it.
So, instead of spending the £350 to £400 on buying a console, you can actually rent and use it for around three years for the same price.
Advertisement
All of this sounds too good to be true, but the PlayStation Flex Raylo page is already live, claiming you can choose from available console and hardware options, select the lease term, sign up in 60 seconds, and get the console delivered the next day.
Sony
A smart expansion strategy?
However, the fact that Raylo conducts a soft credit check does make it sound more realistic.
Although it could be the Japanese giant’s way of getting the inventory rolling, it sounds like a genius idea to me for increasing the console’s adoption by breaking into a previously unexplored audience.
Unfortunately, Sony hasn’t confirmed any such program for the United States or other regions. Would you want the company to release “PlayStation Flex” for you?
The Ugreen NASync DH4300 is a very easy-to-use NAS for beginners who need more capacity than your typical entry-level network storage appliance.
Ugreen NASync DH4300 Plus
We’re getting to a point with cloud storage that in the long run, having a local cloud is a good idea. And, network attached storage (NAS) isn’t just for networking experts anymore. But, it’s easy to underestimate needs. That local cloud basic user could go for entry-level options with two drives, which is fine for most, but when you start thinking about other stuff to do, simply isn’t enough when data storage needs grow. Continue Reading on AppleInsider | Discuss on our Forums
Ring is cancelling a partnership with Flock Safety that would have allowed law enforcement using Flock’s system to request footage from Ring camera owners for voluntary sharing through the Ring Neighbors app. (Ring Photo)
Amazon-owned Ring has canceled its planned partnership with Flock Safety, the license-plate-recognition company, amid broader scrutiny of the potential for home security cameras to be used as tools for neighborhood monitoring and law enforcement.
In a statement Thursday afternoon, Ring said the companies made a “joint decision to cancel the planned integration,” saying it “would require significantly more time and resources than anticipated.” Ring said the integration never launched and no customer videos were shared.
This comes days after Ring’s Super Bowl ad for its AI-powered Search Party feature for finding lost dogs caused a backlash from critics concerned it could be used for surveillance. That was despite Ring’s assurances that the feature gives camera owners full control over whether to share footage with the owners of lost dogs, mirroring its broader privacy policies.
The Flock partnership wasn’t directly related to the Search Party feature, but it was referenced in media coverage as evidence of Ring’s broader collaborations with law enforcement.
The partnership, announced in October, would have allowed local law enforcement agencies using Flock’s platforms to post requests for footage directly in Ring’s Neighbors app during active investigations. Ring users could then choose to share video or ignore the request.
Advertisement
Flock’s technology is used by thousands of police departments. Civil liberties groups including the ACLU had raised concerns that footage could ultimately be accessed by federal agencies.
Ring has said it has no partnership with ICE and does not share video with the agency.
The company’s Community Requests feature, which allows local law enforcement to request footage from nearby Ring users during active investigations, remains in place. Participation is voluntary, allowing users to choose to share footage or ignore the request.
Rival home security company Nest, owned by Google, is separately in the spotlight in the Nancy Guthrie disappearance case, where investigators this week recovered video from a Nest doorbell camera that had no active subscription, with Google’s help.
Advertisement
Speaking with CBS News on Thursday afternoon, Ring founder Jamie Siminoff noted that Ring’s system is designed so that video data doesn’t exist for users without a subscription, distinguishing Ring’s approach from what happened in the Guthrie case.
Siminoff emphasized that the Community Requests allow police to ask for footage from Ring customers in a “privacy-protected way,” and said the system has already aided investigations, including a shooting near Brown University in December.
He also used the appearance to defend Search Party, saying the feature was built “privacy first” and comparing it to finding a lost dog in your backyard and calling the number on its collar.
We spend hours testing every product or service we review, so you can be sure you’re buying the best. Find out more about how we test.
I’m not going to get to have the Acer Veriton GN100 for long, so this is more of a hands-on discussion than an actual review.
My first reaction, out of the box, to the Acer Veriton GN100 is that it all seems remarkably familiar. An elegant mini-PC style case with a car-grill aesthetic, a selection of USB-C ports alongside a 10GbE LAN port and the mercurial NVIDIA ConnectX-7 SmartNIC.
Advertisement
While it’s physically a little smaller, the ports on this machine are identical to those on the ASUS Ascent GX10, as both brands followed Nvidia’s Blackwell system plan exactly.
The only significant difference is that where the Asus provided access to the single M.2 NVMe drive that was installed, the Acer Veriton GN100 is an entirely sealed unit, where whatever storage it has can’t be replaced or upgraded.
Inside is the same Nvidia DGX Spark Personal AI Supercomputer built around the ARM v9.2-A CPU and Blackwell GPU integrated silicon. This, when combined with 128GB of LPDDR5 memory and 4TB of NVMe storage, is collectively called the GB10 platform.
This is a remarkably powerful platform that has uses in data science, medical image analysis, robotics and AI model development. To be clear, this isn’t a Windows PC, and an understanding of Linux is required to use it.
Advertisement
As the specification suggests, this isn’t an inexpensive item, starting at $2999 for the US-supplied hardware, but £3999.99 in Europe. For those who want a highly compact and efficient development environment, especially for AI, the Acer Veriton GN100 is an option, but it isn’t the only machine available using the same platform.
Sign up to the TechRadar Pro newsletter to get all the top news, opinion, features and guidance your business needs to succeed!
Acer Veriton GN100: Price and availability
How much does it cost? From $3000, £4000
When is it out? Available now
Where can you get it? Available from Acer and online retailers
To avoid any confusion about specifications, Acer decided there would be only one SKU of the Veriton GN100, with 128GB of LPDDR5 and 4TB of storage.
Inexplicably, it costs £3999.99 direct from Acer in the UK, but only $2999.99 from Acer in the US. Why do we pay 82% more for an identical part when the UK doesn’t tariff Taiwanese goods, Acer?
Advertisement
Also, this product doesn’t seem to be available elsewhere, so finding it cheaper on Amazon, for example, isn’t currently possible.
The alternatives built around the same platform are the ‘founders edition’ Nvidia DGX Spark Personal AI Supercomputer, ASUS Ascent GX10, Gigabyte AI TOP ATOM Desktop Supercomputer, and MSI EdgeXpert Desktop AI Supercomputer.
The Nvidia DGX Spark Personal AI Supercomputer, as the originator modestly calls it, undercuts the Veriton GN100 in the UK and costs £ 3699.98 for a system with 128GB of RAM and 4TB of storage. But, it’s more expensive for US customers, costing $3999 on Amazon.com.
The ASUS Ascent GX10 price on Amazon.com is $3088.94 for the 1TB storage SKU (GX10-GG0015BN), and $4,149.99 for the 4TB storage model (GX10-GG0016BN).
Advertisement
Even with the current price of M.2 modules, that is a remarkable price hike for the extra storage capacity.
For UK customers, the 1TB ASUS Ascent GX10 model price is £3713.02, but I found it via online retailer SCAN for a tempting £2799.98. SCAN also carries a 2TB option for $3199.99 and the 4TB model for £3638.99.
The Gigabyte AI TOP ATOM Desktop Supercomputer 4TB model sells for £3479.99 from SCAN in the UK, and can be found on Amazon.com for $3999.
And the final model with the same spec as most is the MSI EdgeXpert Desktop AI Supercomputer, selling for £3,598.99 from SCAN in the UK, and $3998.01 on Amazon.com for US customers.
Advertisement
The conclusion is that the US price is difficult to beat, whereas the European pricing is wildly out of what competitors are charging for this technology.
ARM v9.2-A CPU (GB10) (20 ARM cores, 10 Cortex-X925, 10 Corex-A725)
GPU:
NVIDIA Blackwell GPU (GB10, integrated)
Advertisement
RAM:
128 GB LPDDR5x, unified system memory
Storage:
4TB M.2 NVMe PCIe 4.0 SSD storage
Advertisement
Expansion:
N/A
Ports:
3x USB 3.2 Gen 2×2 Type-C, 20Gbps, alternate mode (DisplayPort 2.1)
Advertisement
1x USB 3.2 Gen 2×2 Type-C,with PD in(180W EPR PD3.1 SPEC)
1x HDMI 2.1
1x NVIDIA ConnectX-7 SmartNIC
Networking:
Advertisement
10GbE LAN, AW-EM637 Wi-Fi 7 (Gig+) , Bluetooth 5.4
OS:
Nvidia DGX OS (Ubuntu Linux)
PSU:
Advertisement
48V 5A 240W
Dimensions:
150 x 150 x 50.5 mm
Weight:
Advertisement
1.2kg
Acer Veriton GN100: Design
Oversized NUC
Connect-7 scalability
Zero internal access
(Image credit: Mark Pickavance)
While the GN100 looks like an oversized NUC mini PC, at 1.2kg, it’s heavy, although it is lighter than the ASUS Ascent GX10 by over 200g.
In order to drive the monster silicon inside, Acer included a Delta-made PSU that’s rated to 240W over USB-C.
Advertisement
All the ports are on the back of this system, and nothing is on the front, not even the power button, other than some visual styling and the Acer logo.
These include, identical to the ASUS Ascent GX10, four USB-C ports, one of which is required for the PSU to connect, a single 10GbE LAN port and a single HDMI 2.1 video out.
This arrangement enables a single monitor using HDMI, but additional ones using the USB 3.2 Gen 2×2 ports in DP Alt mode, although one is exclusively needed to power the unit.
Why Nvidia chose USB 3.2 and not USB4 seems curious, since the models and data processed on this unit will eventually need to make it somewhere else, and the best networking on offer is 10GbE, which equals roughly 900MB/s transfer speeds.
Advertisement
And for those working on the hardware, the lack of any USB-A ports for mice or keyboards looks a bit silly.
However, this hardware is intended to be used ‘headless’ using a remote console, so perhaps that isn’t an issue in the greater scheme of things.
Where this design sheds any resemblance to PC hardware is with the inclusion of a ConnectX-7 Smart NIC, a technology acquired by Nvidia when it bought Mellanox Technologies Ltd, an Israeli-American multinational supplier of computer networking products based on InfiniBand and Ethernet.
In this context, ConnectX-7 is like those annoying cables that Nvidia used to make video cards work collectively, when they cared about video cards. Except that the capacity amount of bandwidth that can travel over ConnectX-7 is substantially more.
Advertisement
The port has two receptacles, with each capable of 100GbE, allowing 200GbE to flow between the GN100 and another, doubling the number of AI parameters from 200 billion in a single machine to 400 billion when buddied up to another.
(Image credit: Mark Pickavance)
Acer Veriton GN100: Features
ARM 20-core CPU
Grace Blackwell GB10
AI platforms compared
The Nvidia GB10 Grace Blackwell Superchip marks a notable advancement in AI hardware, created through a partnership between Nvidia and ARM. It arises from the growing need for specialised computing platforms to keep pace with the rapid development and deployment of artificial intelligence models. Unlike a typical PC, the GB10 is designed around the ARM v9.2-A architecture, incorporating 20 ARM cores (10 Cortex-X925 and 10 Cortex-A725). This reflects a wider industry move towards ARM-based options, which are more power-efficient than PC processors, and potentially more scalable for AI tasks.
The capabilities of the GB10 are impressive. It combines a robust Nvidia Blackwell GPU with the ARM CPU, achieving up to a petaFLOP of AI performance with FP4 precision. This level of power is especially suitable for training and inference of large language models and diffusion models, which are fundamental to much of today’s generative AI. The system is further supported by 128GB of unified LPDDR5x memory, enabling it to handle demanding AI tasks efficiently.
The caveat to all this power and memory is that PC architectures aren’t designed to exploit them effectively, and Microsoft Windows memory management has long been an issue.
Advertisement
Therefore, to be efficient and communicate effectively with other nodes, the GB10 needs Ubuntu Linux, modified with NVIDIA’s DGX OS, to harness the platform’s power and handle multi-node communications.
(Image credit: Mark Pickavance)
As I already mentioned, the GB10 delivers up to 1 petaFLOP at FP4 precision, ideal for quantised AI workloads. But that is still less than the multi-petaFLOP performance of NVIDIA’s flagship data centre chips, the Blackwell B200 or GB200.
However, where it goes toe-to-toe is in respect of power efficiency, since this node only consumes around 140W, which is much less than the Blackwell B200, which can consume between 1000W and 1200W per GPU. The GB200 combines two B200 chips and a Grace CPU, and the power demand can bloom to 2,700W. Although these systems might offer up to 20 Petabytes of performance, at around 19 times the power.
The balance here is that the GN100 can sit on your desk without needing any special services or environment, whereas the datacentre hardware needs a specialist location and services to ensure it doesn’t overheat or cause the local electricity network to fail.
Advertisement
In this respect, the GN100 and its counterparts represent the more realistic end of the AI wedge, but how useful they can be is dependent on what you are attempting to do, and if this much power is enough for your purposes.
Acer Veriton GN100: AI Reality Check
(Image credit: Mark Pickavance)
In my prior Asus GX10 coverage, I talked at length about AI, and how there are lots of people making a bet that it is the next big thing, and others who are much more critical of the technology and how it’s developing.
I’m not going to rehash the obvious flaws of AI, or the lack of a path to address all of those, but I would strongly recommend researching before starting any AI endeavour and creating expectations that either can’t be met with current technology, or the power in this physically small computer.
Advertisement
What I can say is that recent AI releases have substantially improved over previous generations, but access to these advanced models, like ChatGPT 5.3 Codex and Claude 4.0, is ringfenced for paid subscribers using the Cloud.
Obviously, the beauty of a device like the GN100 is that you can download these models and run them on your own hardware, even if getting the most out of them requires them to be connected to the Internet to source information.
For those interested, running GPT-5.3-Codex on this hardware requires you to install Tailscale and a local inference engine like Ollama, pull the codex to the GN100 using the appropriate commands. Then you can open up an Open WebUI from another system, ideally, and use the model.
For anyone familiar with Linux, none of this is especially taxing. But to make it even easier, the ChatGPT team (or is that the AI?) has made a Codex App that does most of the legwork for you.
Advertisement
If you want to try something else, some models come in pre-prepared Docker Containers that can simply be installed and executed, making deployment remarkably straightforward.
(Image credit: Mark Pickavance)
When you first power the system up, you are presented with a web interface created for the Nvidia DGX Spark, and install VS Code, the DGX Dashboard with JupyterLab, Open WebUI with Ollama and Comfy UI.
To be clear, this isn’t like a Windows application install. You are given instructions as to the commands you need to execute, which install the tools and libraries that are needed. Those who don’t use Linux every day will find it a challenge, but eventually, even I managed to get almost everything working, at least enough to load models and create some output. Image generation was especially impressive, although some of the deep thinking models aren’t that responsive if you ask them something genuinely challenging.
Some people might assume that because the code for these models is being run locally on the GN100, the cost of a subscriber model should be cheaper. But I’ve not noticed that so far, you just get better performance and save the creators’ electricity bill. What running a model on this hardware gives you is that you own the model, it can’t be removed from you, and there is the potential for you control the model, customising it in a specific and personal way.
Advertisement
For those exploring AI in a serious way its necessary to use the latest models, and that often has a cost implication, even if you own the hardware platform.
What you certainly don’t want to do is install some free model from a couple of years ago, and then be disappointed with the results. The steepness of the curve of development on models is extraordinary, and even versions of the best ones from six months ago have been overtaken by the latest releases.
For those working in this area, using modern AI is like trying to get onto a bus when they don’t stop or even slow down to allow passengers on or off. Being aware of where the model of interest is, and when it’s been overtaken, is critical to not being completely out of date before the project is completed.
Advertisement
Acer Veriton GN100: Early verdict
(Image credit: Mark Pickavance)
NVIDIA decided to ship its Grace Blackwell technology in an entry-level form and created a blueprint for that in their NVIDIA DGX Spark Personal AI Supercomputer, its partners are delivering their versions, like the Acer Veriton GN100.
Other than the outer case and a few other small choices, it’s debatable how much variation we’re likely to see between these machines. It’s not like the GPUs, where the partners get to design variations and even tweak the founders’ editions for better performance.
Maybe if these become massively popular, then we might see more variation, like combining two systems in a single box, or blending the technology with DAS storage. But for now, this is where we are.
That said, the Acer version is perfectly serviceable, but when the specifications are so close its mostly about price. For Americans paying $2999 for the GN100, it’s probably one of the cheapest options, and for those in Europe, oddly, it’s one of the more expensive. Perhaps Acer can fix that for Europeans, but given the rising cost of RAM, it’s more likely the USA will have to pay more.
Advertisement
The one weakness of this design is the lack of access to the SSD, and if that’s a deal breaker for you, some other machines do have that capability.
As ever, Acer has delivered a workable solution for demanding computing tasks, but what this brand can’t guarantee is the skills needed to make the most from their platform. Buyers need to appreciate that while the hardware offers more than ten times the AI processing of a high-end PC, making the most of what it can do requires a particular skill set.
For more compact computing, see our guide to the best mini PCs you can buy
Telecom operators planning aggressive fiber and fixed wireless broadband rollouts in 2026 face a serious supply problem — DRAM and NAND memory prices for consumer applications have surged more than 600% over the past year as higher-margin AI server segments absorb available capacity, according to Counterpoint Research.
Routers, gateways and set-top boxes have been hit hardest, far worse than smartphones; prices for “consumer memory” used in broadband equipment jumped nearly 7x over the last nine months, compared to 3x for mobile memory. Memory now makes up more than 20% of the bill of materials in low-to-mid-end routers, up from around 3% a year ago. Counterpoint expects prices to keep rising through at least June 2026. Telcos that were also looking to push AI-enabled customer premises equipment — requiring even more compute and memory content — face additional headwinds.