Feed aggregator

This review first appeared in issue 354 of PC Pro.

It’s not unusual for Linux distributions to be available in several builds. Ubuntu has long-term support and cutting-edge editions, Zorin has Core and Pro builds, and openSUSE has Tumbleweed and Leap. Tumbleweed, which we’re reviewing here, is a continuously evolving build comprising the latest stable version of each component, which is updated as and when new code appears. Leap, which is based on SUSE Enterprise Linux, is a stable release that receives periodic updates.

The version that we installed was running the 6.6.3-1 kernel (the very latest build was 6.6.4 at the time of writing), and is available with a choice of desktop managers, including Gnome, Xfce, Cinnamon, Mate and KDE Plasma 5, the latter of which tops the installer’s list of options. That’s therefore what we chose for our tests. In our installation, this sat on top of the X11 Window System rather than Wayland.

As well as the regular 64-bit and 32-bit builds for desktop use, there are alternative builds for ARM, IBM Z family mainframes, RISC and PowerPC. It will also run on Raspberry Pi.

The installer is logical, but not the most friendly on test. Setting up Wi-Fi requires that you know which security protocol your network uses, and we failed on our first three attempts to boot into openSUSE at the end of the process. Despite the installer apparently completing, our test machine still attempted to start the OS we had previously tested. We solved this by forcing the installer to delete all partitions on our internal drive even if it thought that wasn’t necessary.

Once it was up and running, however, everything was far more logical, and there were only ten updates ready to download, reflecting the fact that the installer was still fresh, having been downloaded just the day before. A further 11 appeared later in the day.

The installer is logical, but not the most friendly on test (Image credit: Future)

It’s not only core components such as SAMBA, git and the kernel that receive rolling updates in Tumbleweed: bundled applications are likewise cutting edge. Thus, at the time of writing we found LibreOffice 7.6.3.1 and Firefox 120 pre-installed. Email is handled by Kmail, but if you prefer Thunderbird, the version accessible through Discover, openSUSE’s software manager, is version 115.5. The versions of GIMP and Inkscape available through the manager, both of which are options rather than pre-installed software, match the latest releases available through their respective sites.

KDE Plasma should be familiar to anyone switching from Windows, as it features a traditional taskbar and menu. Commonly used applications can be pinned to the bar for easy access, and menu options are filtered as you type to reduce the number of matching applications.

Plasma lets you add secondary toolbars, and also features widgets, with a generous selection pre-installed but dormant. We saw this implementation in Nitrux, which also uses KDE Plasma, although in a significantly tailored form. For our money, the version seen here, where window controls remain attached to the windows themselves, feels more logical.

The comprehensive settings app is supplemented by YaST (Yet another Setup Tool), which provides password protection for more sensitive options, such as configuring the boot loader, administering printers and managing users. Tumbleweed scored 1,167 in the single-core Geekbench tests and 3,241 in the multicore benchmark. This puts it broadly in line with the other distributions on test, suggesting that deployment decisions would be better based on factors other than performance. openSUSE is one of the oldest Linux distributions still being actively developed but, with Tumbleweed, you’ll always be running one of the newest builds available. This is its primary appeal, since other factors such as the choice of desktop managers, and pre-installed applications, are common to many rivals. Although we initially experienced some difficulty with installation, your mileage will likely vary and, once up and running, it was regularly updated – as promised – and as stable as any other distribution on test.

If your preference is for KDE, rather than Gnome, this would be our pick of the bunch.

We've also ranked the best project management software.

This review first appeared in issue 354 of PC Pro.

On first encounter, Nitrux 3.2 is a moody beast. The default UI is dark, file management windows have black backgrounds and apps are dimmed when they lose focus. Yet somehow, even if you’re not a fan of dark modes (as this writer isn’t), it works.

You can switch between windows for some real showmanship: apps swoop around each other to make space, and where two windows sit side by side, they part to let hidden windows through. The default font is skinny but easy on the eye, and while the launcher is too angular for our tastes, that’s a matter of personal preference.

Where we think it’s less successful is in detaching the windows buttons – close, maximize and minimize – and putting them up in the screen’s top left corner, where they sit on the end of the menu bar. This is, presumably, why apps without focus are dimmed: so you know which one you’re going to close or enlarge when you click them. A side effect is that when switching between a browser and word processor, if you’re referring to the former while writing in the latter, your reference material will frequently change brightness.

Elsewhere, the file manager (called Index) has a grid of quick access buttons for common folders, such as Music, Documents and Pictures – but when they all sport the same icon and there’s no label, as was our experience with the default theme, you have to hover over each one and wait for its name to appear in a tooltip.

Buttons in the file manager provide access to folders, but they’re not labelled (Image credit: Future)

If you don’t get on with the default desktop, there are other themes to choose from, including some with more traditional window controls.

The Calamares installer is logical and offers plenty of guidance, such as the requirements for a secure password. Once up and running, you might not recognise a huge number of pre-installed applications other than Firefox, and there’s no office suite ready for use, although LibreOffice is available through the NX Software Center. So are staples such as Thunderbird, GIMP and Inkscape, which are likewise absent. You’ll find placeholders for Steam, Waydroid and Bottles, with the latter two providing layers for Android and Windows apps respectively.

However, it didn’t detect the two printers on our network until we’d entered their IP addresses.

Behind the scenes, this is a Debian-based distribution using the 6.6.2-1 kernel and a modified version of KDE Plasma to render the NX Desktop. There’s a generous selection of desktop widgets, and a configurable button on the dock, which you can use to run scripts or activate an applet.

The root directory is unwritable by default which, says the developer, strengthens security and minimizes points of failure. It should also mean that if an update fails your machine remains bootable. Other security measures have been implemented elsewhere, such as disabling core dumps, implementing strict password expiration and deactivating the root account. There’s a built-in backup tool – Kup – that can be set to kick in after you’ve been using your computer for a set period, which is an unusual and useful option. The X11 display protocol has been dropped in favour of Wayland, providing a shorter and more secure route between application and display.

One happy benefit of the switch to Wayland, which is taking place across most distributions, is a probable speed bump, and certainly Nitrux feels fast in everyday use. This was particularly evident when we were browsing the web, with pages seeming to arrive more quickly on our domestic broadband connection than they do in other Linux distributions or on our Windows box. We suspected that our broadband had been upgraded (and checked), but that wasn’t the case.

There’s a lot of interesting technology in this distribution, and it’s refreshing to encounter such an individual interface. Going down this road is a brave and largely successful move that has garnered significant praise, including when we named it the best Linux distro for new users.

However, with this review written, we’re now looking forward to returning to a more conventional environment.

We also rated the best small business software.

This review first appeared in issue 354 of PC Pro.

A key player in the Zoho software and services empire, ManageEngine offers a wide range of IT management solutions. The OpManager product family delivers network, server and application performance monitoring, and ManageEngine version 12.7 delivers plenty of new features.

There’s a focus on compliance and security, with support for single sign-on using security assertion markup language (SAML) and SNMP 3 with tougher SHA-2 authentication. Root cause analysis (RCA) profiles compare data collected from multiple devices to help with problem remediation, Cisco ACI, IPMI and VPN monitors are present and OpManager integrates with Slack and Microsoft Teams for alert notifications.

ManageEngine’s flexible licensing schemes are good value for SMBs as they’re based on devices, not elements or sensors. The OpManager Standard edition costs £195 per year for ten devices and provides availability, interface and SNMP/WMI network monitoring services, with the Professional edition on review adding features such as virtualization host monitoring.

Along with perpetual licenses, they’re also all available as Plus versions that include OpManager’s add-on modules at discounted prices. For example, Professional Plus starts at £983 per year for 50 devices and includes network configuration management (NCM) and NetFlow analysis (NFA), along with firewall and application management.

Virtual host monitoring is included in the Professional edition (Image credit: Future)

Whichever edition you choose, you won’t wait long to see it in action: it took us 20 minutes to install it on a Windows Server 2022 host and run an initial scan of the lab’s IP subnet. For device identification and classification, OpManager is a veritable sleuth as it’s endowed with over 11,000 device and 56,000 vendor templates.

It had no trouble figuring out what our lab devices were, correctly identifying all our Windows servers and workstations, routers, switches, NAS appliances, VMware and Hyper-V virtualization hosts and printers. Adaptive alert thresholds are assigned to devices where OpManager uses machine learning algorithms to adjust them for you, and they can be linked to a range of actions using the integral drag-and-drop workflow builder.

The OpManager web console and its multitude of dashboards is capable of presenting a lot of information about your network, with the handy heatmap widget showing a grid of colored blocks representing each device and their status. Detailed custom dashboards are created by choosing from over 200 available widgets, and network operations center views can present support departments with the big picture.

Dashboards can be customized to suit, and mobile monitoring apps are included (Image credit: Future)

Virtual monitoring is extensive, as OpManager discovered our VMware ESXi and Hyper-V hosts and provided views of system and VM utilization, guest OSes and datastore usage. Add the NetFlow module and you can view all common flows and use the free NetFlow Generator utility to translate raw packets into NetFlow data.

OpManager’s RCA profiles are similar to the SolarWinds NPM PerfStack feature and are just as easy to create. The list of available metrics is very basic, but you can choose devices and monitored items such as CPU and memory usage, add them to your RCA dashboard and view them all on one screen to help identify complex problems.

The Application Performance Management (APM) plugin is a worthy addition as it monitors and reports on a huge range of apps. We used it to check on our AWS account, where it provided details on service availability, EC2 instance and S3 bucket status, plus monthly billing charges and forecasts.

It may take time to get to grips with its myriad components, but ManageEngine’s OpManager is a powerful monitoring solution with a lot to say about your network. The web console is well designed and easily customized to suit, while SMBs will approve of OpManager’s affordable licensing plans.

We've also ranked the best fast PCs and workstations.

This review first appeared in issue 354 of PC Pro.

Many Debian-based distros, Linux Mint included, use Ubuntu as their jumping-off point. The latest release – 21.2 Victoria – is based on Ubuntu 22.04 LTS and will continue to be supported until 2027. That means it currently uses the 5.15 LTS kernel, but there’s a bleeding-edge edition, aptly called 21.2 Cinnamon Edge, that uses 6.2 should you prefer.

Where many Ubuntu spin-offs stick with Gnome display manager, as used by Ubuntu itself, Mint’s default is Cinnamon. This started as an offshoot from Gnome 2 when version 3 was released. Version 3 was the first to drop support for the panel (taskbar in Windows parlance) in favor of a purer shell, which likewise demoted widgets and jettisoned desktop icons. Cinnamon puts all of these front and center, which is one of the reasons it’s often held up as an ideal distro for anyone switching from Windows. It also has a Windows-like menu, with applications sorted into categories, and essential tools – such as the browser and settings – kept in view up the side. You can pin your most used apps to the panel so they’re never more than a click away.

Cinnamon is certainly a sturdy, tidy interface, but we think it feels a touch old-fashioned when compared directly with Microsoft’s Fluent UI for Windows 11. If it doesn’t immediately grab you, you can tweak the colors, install a theme, or opt for one of the alternative desktop managers, since Mint is also available with Mate or Xfce. And, while it seems highly unlikely that Ubuntu is going to disappear any time soon, the Mint developers are maintaining a parallel build, LMDE – Linux Mint Debian Edition – based on the Debian codebase, from which development can continue.

Whichever build you choose, you’ll need at least 2GB of memory and 20GB of storage, although 4GB and 100GB are recommended. These are hardly onerous, and the same specs apply to the Cinnamon Edge release.

Getting up and running is a familiar process, which starts with booting into a live installer and stepping through the wizard. However, because of changes to Ubuntu’s shim-signed bootloader, compatibility with Secure Boot was broken for the ISO that was current when we performed our tests. We therefore had to disable Secure Boot to proceed and, unless the ISO has been fixed by the time you install, you may have to do the same. Linux Mint says that it’s working on a fix.

Once up and running, pretty much everything was ready for use. Our monitor was recognized and its native resolution selected, our two network printers were installed, and a range of default applications were in place. Three desklets – effectively Windows-style widgets – for a clock, launcher and digital photo frame are installed but not activated. You can add others from a desklet store, where we’d highly recommend the weather desklet and notepad.

The Software Manager makes it easy to install apps (Image credit: Future)

LibreOffice was ready to roll, and while we can’t say the same for GIMP and VNC, they were available through the software manager, which was set up to work with Flatpak. Thunderbird was pre-installed for handling mail, and Firefox was set up as the default browser.

Hypnotix, Linux Mint’s bespoke IPTV application, is perhaps the most impressive of all the default apps. Its integrated catalogue includes links to more than 1,200 broadcasters, organized by country, with 92 from the UK. Pick a country, then click a channel in the sidebar to tune in. It was quick, efficient and very smooth on our consumer broadband connection. At the time of testing, it wasn’t possible to save favorites, but that may change as work is ongoing for a future release.

Linux Mint with Cinnamon is a pleasant place to spend a lot of time, and while we don’t think Cinnamon feels as dynamic as Windows 11, its familiarity will likely be a significant draw for more cautious switchers (although we would urge anyone in that position to also consider “new kid on the block” Zorin). Hypnotix and the desklets are the star turns in this distribution, and they’re reason enough on their own to make it one for the shortlist.

We've also rated the best lightweight Linux distro.

Lost among the hype of the AI tools and the cloud revolution, one critical question is being overlooked: how many dimensions can your storage system scale?

Most enterprise storage systems are designed to scale in at most two key dimensions — capacity (how much data can be stored) and performance (how quickly data can be accessed). Driven by the complex, dynamic and unpredictable nature of AI workloads and cloud-native applications, there’s been a dramatic shift in the demands placed on storage infrastructure. Storage systems built for an earlier era are now being stretched and strained in ways their designers could never have anticipated. These unprecedented demands can’t be met by simply adding sheer capacity or raw performance.

What does this mean for your organization? To keep pace with emerging workloads and unknown future requirements, your storage infrastructure must scale — without tradeoffs — in any dimension that might be conceivably required.

What is “multidimensional scaling” in data storage?

We’ve observed an uptick in customers who’ve come to us after running into one or more limitations in their outdated, inflexible storage infrastructure. This experience has helped us identify ten distinct dimensions that a storage system must be able to seamlessly scale in order to adapt to whatever the future might throw at it:

• Capacity
• Storage compute
• Applications
• Metadata
• S3 objects
• S3 buckets
• S3 authentications per second
• Throughput
• Objects per second
• Systems management

The ability to scale across these ten dimensions can be referred to as multidimensional scaling.

Future-ready storage: three ways multidimensional scaling powers business growth Technical explanations of each of the ten dimensions can be found in our whitepaper. Instead, let’s explore a few concrete examples of how multidimensional scaling can save your organization time, money, and headaches.

1. Have the flexibility to handle any future demand

The pace of technological change makes it nearly impossible to predict future storage requirements. Workloads that barely existed five years ago — like generative AI and real-time analytics — now dominate IT management discussions. Businesses that commit to rigid storage architectures risk being unprepared for what comes next.

Multidimensional scaling equips organizations with the flexibility to adapt to evolving requirements. Need to onboard a new AI data pipeline with microsecond latencies? Scaling access speeds ensures the infrastructure is ready. Launching a cloud-native application that spans geographies? Scaling authentication requests allows for secure, seamless global access.

This adaptability not only prevents expensive data migration but also empowers businesses to innovate without being constrained by their storage infrastructure.

2. Eliminate storage silos for seamless operations

Data silos have long been the bane of IT teams. They fragment workflows, increase costs, and make it nearly impossible to gain a unified view of your data. With multidimensional scaling, businesses can consolidate their cloud storage into a single, unified platform that handles diverse workloads — from massive AI datasets to high-frequency transactional data.

For example, scalable throughput ensures high-definition video streaming platforms can deliver seamless content playback to millions of users without requiring separate systems for ingesting and delivering media. Meanwhile, scaling the number of supported applications means businesses can run multiple workloads — from data analytics to machine learning — on a single storage solution without bottlenecks. The result? A unified system that reduces complexity and operational overhead.

3. Minimize downtime through predictable scaling

Unplanned downtime is the enemy of productivity and customer satisfaction. Traditional storage systems, which often require manual intervention to expand capacity or support increased traffic, introduce risk every time they hit a limit. Multidimensional scaling sidesteps this issue by ensuring storage systems can grow seamlessly in all directions — whether it’s handling higher transaction rates, supporting more simultaneous users, or accommodating ever-growing datasets.

For example, consider an ecommerce platform during a peak shopping event like Black Friday. With the ability to scale in transactions per second, the platform avoids crippling bottlenecks, ensuring smooth and responsive experiences for millions of shoppers. Likewise, metadata scaling supports the explosion of individual product entries and customer profiles without degrading performance. By adopting a solution designed for such scenarios, businesses can keep operations running smoothly even under extreme demand.

The future of storage is multidimensional: Is your system ready to scale?

Storage may not always be in the spotlight, but it’s the foundation of a winning IT strategy. The ability to scale across all critical dimensions turns storage from a reactive, operational necessity into a powerful driver of growth. By eliminating silos, minimizing downtime, and future-proofing against uncertainty, businesses can unlock new opportunities and avoid the pitfalls of a rigid infrastructure.

The demands of AI, cloud computing, and media-rich workloads are accelerating — your storage must keep up. The organizations that adopt multidimensional scaling today will be the ones best equipped to thrive tomorrow and benefit from long-term sustainability.

So, how many dimensions can your storage system scale?

If you don’t know the answer, it’s time to find out.

We've listed the best business cloud storage.

This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro

Here are the answers for The New York Times Mini Crossword for March 7.

A buyout to take the drugstore chain private would give it more flexibility to make changes to improve its business without worrying about Wall Street's reaction.

(Image credit: Gene J. Puskar)

The decision would allow Yoon to stand his criminal trial while not being physically detained. The court is expected to rule soon on whether to uphold his impeachment or remove him from office.

(Image credit: Ahn Young-joon)

From personalized gifts to business essentials, WIRED can help you save with our selection of VistaPrint promo codes.

Check out our deals for Nike this month, including up to 40% off select styles.

Getting an interview for a job can be exhilarating if you've spent a lot of time sending resumes and cover letters out, but that excitement is often matched by trepidation as you now have to convey why you're the one for a role to an actual human, whether in person or over a video call. But the days of manually hunting for the most common interview questions, awkwardly practicing in front of mirrors, and guilting friends and family into practicing may be over, thanks to ChatGPT.

I've used ChatGPT for everything from dinner ideas and children's entertainment to New Year's Resolutions and many other tasks. So, why not see what it can do for your job interview preparation?

Here are some of the best ways to use ChatGPT to help you nail that interview for a job.

Research

One of the first things you'll want to do is research the company you're interviewing for. Yes, you know enough about them to apply for the job, but more information is always a good idea. While that usually has meant filling tabs with articles and interviews bout a company, ChatGPT can streamline the whole thing by summarizing data in as much detail as you want. You can pull together the company's key values, recent news, and the competitive landscape for their industry into digestible nuggets.

For instance, you might prompt ChatGPT with "Give me a concise overview of Patagonia's mission and recent sustainability initiatives," or "Summarize recent developments at Airbnb relevant to a content strategist role."

Come up with questions

Next, consider what questions you might face. Think about your industry and role. Sure, you can resort to internet searches and write down questions in the same ballpark, but the more specific you are, the better you'll prepare. ChatGPT is great at zeroing in on hyper-specific questions if you let it.

For example, you could ask: "Give me interview questions commonly asked for UX design roles in tech startups," or "What kinds of strategic questions might come up during an interview for a product manager at Amazon?"

Instead of generic advice, ChatGPT provides nuanced, detailed lists of questions and can even ask you for more details as it puts together lists to practice with that are relevant to real-world scenarios you might face.

Mock Interviews

With all of this in place, you can start a major aspect of job interview prep with ChatGPT running mock interviews. Usually, you'd beg a friend or mentor to take pity on you and spend a half-hour pretending to be your future boss. The problem? Friends might go too easy on you (unhelpful) or overdo the tough-love approach (also unhelpful but emotionally scarring).

ChatGPT struck a comfortable middle ground, providing realistic questions without making me feel like I'd botched my entire career before it even started. If you want to try it, use a specific prompt for a company, like "Pretend you're interviewing me for a senior programming role at Spotify," or angle it for the kind of questions you might expect, like, "Give me a mock behavioral interview focusing on managing difficult colleagues."

Refine and improve

The mock interview doesn't mean your work is done. Answering questions confidently is one thing; answering them concisely and persuasively is another. Traditionally, refining your responses means scribbling notes and potentially becoming a flat-toned robot in what you say. ChatGPT offers a new way to improve your answers without losing your voice. Think about where your answers need aid and ask ChatGPT for help.

For example, you could prompt the AI with a lame answer like: "My biggest weakness is perfectionism," and then ask, "How can I answer this in a way that sounds genuine and less cliché?" Or you might request, "Make my answer about resolving a team conflict sound less defensive and more collaborative."

Interview AI

ChatGPT can be great for aiding in job interview preparation, but there are a few caveats worth mentioning. While impressive, it can't quite replicate human intuition or emotional intelligence. Sometimes, its suggestions may veer into overly robotic territory, so it might be better to think of its advice as a starting point rather than absolute truth.

Still, when so much is riding on a conversation, the more research, prep, and practice you can do, the better. ChatGPT might not guarantee your dream job, but it can at least help you feel confident and ready to convince the interviewer that you're the one they've been hunting for all this time.

You might also like

• Our predictions for AI in 2025 – what next for ChatGPT, Apple Intelligence and more
• If Sam Altman thinks OpenAI needs to raise ChatGPT prices, I might have to cancel my Plus membership
• Top 3 things you have to try with the new ChatGPT search

• Apple's new 13-inch MacBook Air with M4 starts at $999 or $1,199 for the 15-inch
• The previous-generation MacBook Air with M3 is discontinued
• The even earlier M2 MacBook Air, which previously started at $999, will live on via other retailers

New Apple devices typically mean the previous generation drops to a lower price point or is discontinued altogether. However, with the Mac lineup – especially the MacBook – since Apple Silicon hit the scene, some of the older models have a way of sticking around.

With the launch of the new 13-inch and 15-inch MacBook Air with M4, Apple is discontinuing the 13-inch and 15-inch M3 MacBook Air. Before the launch, there was some talk of the 13-inch version of that laptop sticking around, but Apple’s overall price decrease is causing waves as that model is discontinued.

Further, since the starting price of the 13-inch MacBook Air with M4 is $999 in the United States, Apple is taking the 13-inch MacBook Air with M2 off its virtual and physical store shelves.

But here’s where things get interesting. Like the M1 MacBook Air – which is still an incredible deal from retailers like Walmart – the M2 MacBook Air will be available at select authorized retailers – aka other purchasing channels – in select countries, including the United States.

Furthermore, it will be cheaper than the starting cost of the M4 MacBook Air.

(Image credit: Future / Lance Ulanoff)

This means you can get a modern MacBook Air design with M2 speeds and other tricks for an excellent price. You’ll just need to be okay with no Sky Blue, but hey, you can still get it in Space Gray.

In fact, as of this story publishing, Amazon in the United States has the 13-inch MacBook Air with M2 chip, 16GB of RAM, and 256GB of solid-state storage in Midnight for just $699 – a sweet 30% off the previous MSRP of $999. Plus, you’ll get free fast shipping if you're a Prime member.

If you need Starlight or Space Gray that same configuration is $799, which is still a deal but not as big of a discount. Similarly, the Silver version is a lot more expensive at $924, only an 8% discount.

While Apple hasn’t announced or confirmed the price of what the M2 MacBook Air will normally be at through these channels, the M1 MacBook Air can give us a good idea. Since its Walmart exclusivity, it’s regularly had a list price of $699 but is nearly always discounted additionally. Right now, in Silver, Space Gray, or Gold, it’s just $629 for the M1 chip, 8GB of RAM, and 256GB of solid-state storage.

So if you don't need the latest chip from Apple, that fancy new shade of Sky Blue (though it is lovely), or a CenterStage capable camera, the M2 should definitely at least be considered. Especially since it comes standard with 16GB of RAM, which is a lot better than the previous 8GB.

You can see our review of the 13-inch MacBook Air with M2 here, and check out our first look at the new 13-inch and 15-inch MacBook Air with M4 here.

You might also like

• New foldable iPhone rumors predict Apple’s bold plans – here are 6 things to expect, from cameras to launch date
• Apple levels up the Mac Studio with the M4 Max and unveils its most powerful chip ever, the M3 Ultra
• Apple updates iPad Air with powerful M3 chip and pairs it with Pro-level Magic Keyboard

Nearly two months after an explosion sent flaming debris raining down on the Turks and Caicos, SpaceX launched another mammoth Starship rocket but lost contact minutes into the test flight.

(Image credit: Eric Gay)

More than 5,600 artists signed an open letter protesting the auction, saying that the works used AI models that are trained on copyrighted work.

The lawsuit pushes back against a recent executive order from President Donald Trump that asks funding applicants to agree not to promote "gender ideology" in their work.

(Image credit: Marion Aguas)

Apple is the only major phone manufacturer without a foldable device.

With more compatibility across Windows and Android, Motorola's Smart Connect is providing seamless cross-device integration akin to Apple's.

The president's message to his Cabinet secretaries comes amid continuing questions over the role of billionaire adviser Elon Musk in the drastic reshaping of the federal government.

(Image credit: Alex Wong)

OpenAI's release of GPT-4.5 for ChatGPT Plus subscribers this week had me immediately keen to try it out, though with some skepticism. There had been reports of OpenAI (and other developers) struggling to make the big improvements to their models we've seen before. And the current standard model GPT-4o, is pretty good in most cases. And if you want a comprehensive report, there's the Deep Research feature.

OpenAI claims that GPT-4.5 has a high emotional intelligence and nuanced understanding of what you say to it. The company's description painted a picture of both models as reliable friends, but that GPT-4.5 would be the one you'd expect a book of poetry from on your birthday. So, I decided to test 4.5 against 4o with a few prompts that any casual ChatGPT user might deploy.

AI poetry

(Image credit: ChatGPT Screenshots)

With that analogy in my head, I decided to start with a poetic challenge. I asked both models, "Can you write me a short poem about a rainy afternoon in New York City and make an image for it?" It seemed fair since looking out a rain-streaked window at a busy city can bring out the poet in most people.

GPT-4.5 is on the left and GPT-4o is on the right. They are amazingly similar. I personally think GPT-4.5 did a slightly better job with similar ideas. It's evocative of not just the look of rain but the feeling of gray skies, puddles, and traffic among the raindrops. In a blind test of three random friends, two out of three chose the same, with the third saying they just preferred the rhyme scheme of GPT-4o.

As for the images, both models used DALL-E 3, but GPT-4.5's looks a lot more realistic. I actually prefer the impressionistic lighting of GPT-4o's attempt, but both get the idea of the poem across pretty well.

Oscar insight

(Image credit: ChatGPT Screenshots)

With the recent Oscars, I went with a fun trivia question with a more complex, subjective follow-up: "Who won the Best Actress Oscar in 2023, and what made their performance stand out?"

Both had the right answer of Michelle Yeoh for "Everything Everywhere All at Once," but GPT-4.5 had a really nice explanation why her performance resonated with viewers. It covered her performance and mentioned how Yeoh was the first Asian to win that Oscar. GPT-4o's answer had a lot of the same beats, but it went with an odd essay and numbered list format that was kind of annoying to read when the question was a simple opinion request. GPT-4.5's answer felt more like how an actual human would answer, albeit one who is very into that movie and Yeoh as a performer.

Easy chef

(Image credit: ChatGPT Screenshots)

Finally, I put the models up against the eternal daily struggle with this prompt: "I'm bored of pasta. Can you suggest an easy dinner recipe that's different but not complicated?"

I don't know what kind of culinary inspiration I was expecting, but GPT-4.5 hit the nail on the head with honey garlic salmon with roasted broccoli. You can see it's a very simple, short recipe, and seems easy and quick to make. GPT-4o went with garlic shrimp and roasted vegetables. That's not a bad idea for a meal (though I personally would need to change the protein), but you can see the preparation alone takes multiple pans to prepare and the ingredient list couldn't even fit in the same space as the entire recipe from GPT-4.5. I asked for easy and not complicated, GPT-4.5 delivered that far better, I would say.

Deep thoughts

GPT 4.5 gives the vibe of recognizing the unstated preferences in a prompt compared to GPT-4o. But, it's very subtle in picking up on those emotional undertones. Without direct comparison, you might never know there's any difference.

GPT-4o is efficient, accurate, and practical in its responses, and its poetry is completely adequate. I wouldn't say those on the free tier using it are missing much from not having access to GPT-4.5, at least right now. Over time, that may change since GPT-4.5 is still a research preview and not fully complete and out. I'm going to keep experimenting with both and see if there are types of prompts where the new model is definitively better, but for now, I'd say don't worry too much about which model you employ, unless you're planning a meal and don't have a lot of time to prepare.

You might also like

• I can get answers from ChatGPT, but Deep Research gives me a whole dissertation I'll almost never need
• I replaced my to-do list with ChatGPT’s Tasks feature and it completely changed the way I plan my life
• Now that ChatGPT Voice Mode is free, is it even worth paying for ChatGPT Plus anymore?

Leaders of the European Union have voted to approve a plan to dramatically increase defense spending. The move is a reaction to the U.S. reducing its support for Ukraine in the war against Russia's invasion. We go to Brussels to understand the ramifications.

And we hear what the series of recent developments in the relationship between the U.S. and Ukraine could mean for peace talks with Russia.

(Image credit: HATIM KAGHAT)