TechRadar News

The excitement about DeepSeek is understandable, but a lot of the reactions I’m seeing feel quite a bit off-base. DeepSeek represents a significant efficiency gain in the large language model (LLM) space, which will have a major impact on the nature and economics of LLM applications. However, it does not signal a fundamental breakthrough in artificial general intelligence (AGI), nor a fundamental shift in the center of gravity of AI innovation. It’s a sudden leap along an expected trajectory rather than a disruptive paradigm shift.

DeepSeek’s impressive achievement mirrors the broader historical pattern of technological progression. In the early 1990s, high-end computer graphics rendering required supercomputers; now, it’s done on smartphones. Face recognition, once an expensive niche application, is now a commodity feature. The same principle applies to large language models (LLMs). The surprise isn’t the nature of the advance, it’s the speed.

For those paying attention to exponential technological growth, this isn’t shocking. The concept of Technological Singularity predicts accelerating change, particularly in areas of automated discovery and invention, like AI. As we approach the Singularity, breakthroughs will seem increasingly rapid. DeepSeek is just one of many moments in this unfolding megatrend.

DeepSeek’s architectural innovations: impressive, but not new

DeepSeek’s main achievement lies in optimizing efficiency rather than redefining AI architecture. Its Mixture of Experts (MoE) model is a novel tweak of a well-established ensemble learning technique that has been used in AI research for years. What DeepSeek did particularly well was refine MoE alongside other efficiency tricks to minimize computational costs:

Parameter efficiency: DeepSeek’s MoE design activates only 37 billion of its 671 billion parameters at a time. This means it requires just 1/18th of the compute power of traditional LLMs.

Reinforcement learning for reasoning: Instead of manual engineering, DeepSeek’s R1 model improves chain-of-thought reasoning via reinforcement learning.

Multi-token training: DeepSeek-V3 can predict multiple pieces of text at once, increasing training efficiency.

These optimizations allow DeepSeek models to be an order of magnitude cheaper than competitors like OpenAI or Anthropic, both for training and inference. This isn’t a trivial feat—it’s a major step toward making high-quality LLMs more accessible. But again, it’s a stellar engineering refinement, not a conceptual leap toward AGI.

The well-known power of open-source

One of DeepSeek’s biggest moves is making its model open-source. This is a stark contrast to the walled-garden strategies of OpenAI, Anthropic and Google – and a nod in the direction of Meta’s Yann LeCun. Open-source AI fosters rapid innovation, broader adoption, and collective improvement. While proprietary models allow firms to capture more direct revenue, DeepSeek’s approach aligns with a more decentralized AI future—one where tools are available to more researchers, companies, and independent developers.

The hedge fund HighFlyer behind DeepSeek knows open-source AI isn’t just about philosophy and doing good for the world; it’s also good business. OpenAI and Anthropic are struggling with balancing research and monetization. DeepSeek’s decision to open-source R1 signals confidence in a different economic model—one based on services, enterprise integration, and scalable hosting. It also gives the global AI community a competitive toolset, reducing the grip of American Big Tech hegemony.

China’s role in the AI race

Some in the West have been taken aback that DeepSeek’s breakthrough came from China. I’m not so surprised. Having spent a decade in China, I’ve witnessed firsthand the scale of investment in AI research, the growing number of PhDs, and the intense focus on making AI both powerful and cost-efficient. This isn’t the first time China has taken a Western innovation and rapidly optimized it for efficiency and scale.

However, rather than viewing this solely as a geopolitical contest, I see it as a step toward a more globally integrated AI landscape. Beneficial AGI is far more likely to emerge from open collaboration than from nationalistic silos. A decentralized, globally distributed AGI development effort—rather than a monopoly by a single country or corporation—gives us a better shot at ensuring AI serves humanity as a whole.

DeepSeek’s broader implications: The future beyond LLMs

The hype around DeepSeek largely centers on its cost efficiency and impact on the LLM market. But now more than ever, we really need to take a step back and consider the bigger picture.

LLMs are not the future of AGI

While transformer-based models can automate economic tasks and integrate into various industries, they lack core AGI capabilities like grounded compositional abstraction and self-directed reasoning.

If AGI emerges within the next decade, it’s unlikely to be purely transformer-based. Alternative architectures—like OpenCog Hyperon and neuromorphic computing—may prove more fundamental to achieving true general intelligence.

The commoditization of LLMs will shift AI investment

DeepSeek’s efficiency gains accelerate the trend of LLMs becoming a commodity. As costs drop, investors may begin looking toward the next frontier of AI innovation.

This could drive funding into AGI architectures beyond transformers, alternative AI hardware (e.g., associative processing units, neuromorphic chips), and decentralized AI networks.

Decentralization will shape AI’s future

The AI landscape is shifting toward decentralized architectures that prioritize privacy, interoperability, and user control. DeepSeek’s efficiency gains make it easier to deploy AI models in decentralized networks, reducing reliance on centralized tech giants.

DeepSeek’s role in the AI Cambrian explosion

DeepSeek represents a major milestone in AI efficiency, but it doesn’t rewrite the fundamental trajectory of AGI development. It’s a sudden acceleration along a predictable curve, not a paradigm shift. Still, its impact on the AI ecosystem is significant:

It pressures incumbents like OpenAI and Anthropic to rethink their business models.

It makes high-quality AI more accessible and affordable.

It signals China’s growing presence in cutting-edge AI development.

It reinforces the inevitability of exponential progress in AI.

Most importantly, DeepSeek’s success should serve as a reminder that AGI development isn’t just about scaling up transformers. If we truly aim to build human-level AGI, we need to go beyond optimizing today’s models and invest in fundamentally new approaches.

The Singularity is coming fast—but if we want it to be beneficial, we must ensure it remains decentralized, global, and open. DeepSeek is not AGI, but it’s an exciting step in the broader dance toward a transformative AI future.

We've featured the best AI phone.

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

• Hostinger launches Horizons, a no-code web app creator
• Duolingo, Notion and Airbnb references as web app startups
• It’s available to all users next month and will save them “weeks or months”

Hostinger is betting big on AI with its launch of a new tool designed to help users build, edit and publish their own web apps without the need for any coding expertise.

Hostinger Horizons has already started rolling out to a select group of customers who have been using the no-code tool to build web apps like a language-learning card game and a time management tool.

The web hosting platform highlighted the versatility of web apps, which offer more interactive and personal experiences compared with traditional websites.

Hostinger Horizons

Hostinger revealed the “vast majority” of its more than 3.5 million customers build traditional sites, highlighting how companies like Duolingo, Notion and Airbnb had initially started off as simple web apps – sites that had previously required greater expertise to build.

“Coding web apps from scratch takes weeks or months, and hiring a developer can cost thousands. For larger projects, both time and costs grow exponentially," noted Chief Product and Technology Officer Giedrius Zakaitis.

To build a web app with Horizons, users will interact with a chatbot which uses AI to create, add and edit components on the site. The chat interface also has support for voice prompts and image uploads, and from launch, it’s set to support more than 80 languages.

Zakaitis added: “Web apps have turned ideas into million-dollar startups, but building one always required coding or hiring a developer. We believe it is time to change the game.”

Once complete, users can publish their web apps to a custom domain, and they can always come back to it to refine and update at any time.

The new product launch comes just over a year after Hostinger added Kodee, a chat assistant that offers users guidance when it comes to building their site.

With four in five of the company’s Website Builder users now utilizing artificial intelligence during their site-building phase, Horizons expands AI’s usefulness to a different type of website.

Hostinger says Horizons will be generally available for new and existing customers beginning next month.

You might also like

• It's finally here: Wix launches long-awaited AI-powered website builder
• Cut the costs with the best free website builders
• We’ve listed the best web hosting providers

• Apple will invest $500bn in the US by 2029 – the equivalent of 12 months’ revenue
• Servers, currently built out of the country, will be manufactured in Houston
• 20,000 new jobs are on the cars, in R&D, silicon, software, AI and ML

Apple has committed to spending $500 billion over the next four years to invest in its facilities and operations across the US, marking its largest ever commitment in the States.

Central to the announcement is Apple’s intention to support US manufacturing – something that President Trump is keen to do as he looks to reduce the country’s reliance on China, which has been a longstanding manufacturing partner for almost all companies.

Michigan, Texas, California, Arizona, Nevada, Iowa, Oregon, North Carolina and Washington will all benefit from the Cupertino giant’s cash.

Apple invests half a trillion into the US economy

Over the next four years, the $3.69 trillion company (and currently the world’s most valuable company) is set to invest half a trillion dollars into the States, which equates to roughly one year’s revenue given that the company posted a 4% year-over-year rise in quarterly revenue to $124.3 billion in the three months ending December 28, 2024.

CEO Tim Cook said: “We are bullish on the future of American innovation, and we’re proud to build on our long-standing U.S. investments with this $500 billion commitment to our country’s future.”

Key to the announcement is a 250,000-square-foot server manufacturing facility in Houston, Texas, where Apple Intelligence-supporting servers are set to be built. Until now, servers have been manufactured outside of the US.

Data center capacity in North Carolina, Iowa, Oregon, Arizona and Nevada will also follow as the iPhone maker looks to power its AI.

Apple will also double its US Advanced Manufacturing Fund from $5 billion to $10 billion and increase R&D spend, which it says is important given the launch of its new C1 cellular modem – its first in-house effort.

Looking ahead, Apple says it will focus on R&D, silicon engineering, software development, and AI and machine learning roles over the next four years, hiring an estimated 20,000 new workers on top of the 2.9 million jobs it currently supports in the US.

You might also like

• Check out the best AI tools and best AI writers
• Meta reveals huge new 50,000km subsea cable project
• Access the most powerful chips with the best cloud computing services

• CISA adds Craft CMS bug to its KEV catalog
• The bug was found in Craft CMS versions 4 and 5
• It allows for remote code execution

The US Government's Cybersecurity and Infrastructure Security Agency (CISA) has added a new bug in Craft CMS versions 4 and 5 to its Known Exploited Vulnerabilities (KEV) catalog, ringing the alarm for abuse in the wild.

The vulnerability is a remote code execution (RCE) flaw tracked as CVE-2025-23209, but we don't know too many details about it, other than the fact exploitation is not that straightforward.

To abuse the bug, a threat actor first needs to have the installation security key, a cryptographic key that secures things like user authentication tokens, session cookies, database values, and more.

Decrypting sensitive data

Threat actors with possession of this bug can decrypt sensitive data, generate fake authentication tokens, or run malicious code from a distance.

Being added to KEV means that CISA has evidence someone is abusing the flaw in real-life attacks. However, the agency did not detail the attacks, so we don’t know who the threat actors are, or who the victims are. The deadline to patch the CMS is March 13, 2025. Admins should look for versions 5.5.8 and 4.13.8.

Admins suspecting compromise should delete old keys contained in the '.env' files and generate new ones using php craft setup/security-key command. They should also be careful not to destroy previously encrypted data, since the new key cannot grant access to it.

Craft CMS is a content management system designed for developers and content creators. The company advertises it as a customizable and intuitive platform with powerful templating, clean control panel, and robust content modeling.

There are many ways in which cybercriminals can abuse flawed content management systems. For example, they can redirect the visitors to a malicious phishing page, stealing their sensitive data in the process. They can serve them malicious ads or, in more extreme cases, drop malware to their computers.

You might also like

• The many moving parts of business email compromise
• We've rounded up the best password managers
• Take a look at our guide to the best authenticator app

• The Scuf Valor Pro is available to order now for $99.99 / £99.99
• It seems like a great value Xbox pro controller with lots of top features...
• ... but it doesn't have wireless connectivity

High-end peripheral brand Scuf has unveiled the Scuf Valor Pro, a relatively affordable Xbox pro controller that looks like a surprisingly comprehensive package for its price.

For $99.99 / £99.99 (which is whopping $80 / £60 less than the Xbox Elite Wireless Controller Series 2) you're getting a premium gamepad that's compatible with Xbox consoles and PC. It's the first Scuf Xbox controller to feature the highly desirable Hall effect thumbsticks, which are more long-lasting and help prevent thumbstick drift.

It boasts an all-new audio control system that allows you to tweak your headset audio via two little dials directly on the gamepad. You also get what the company is calling 'contoured bumpers' - a larger bumper shape that wraps around the top of the controller for faster inputs.

This is on top of a barrage of the expected Pro controller features, such as four remappable rear paddles, the ability to save up to three unique control profiles, instant trigger locks, and rubberized grips. While no carrying case is included in the box, you do get two additional thumbsticks to further customize your controller.

On paper, the Scuf Valor Pro looks like a big upgrade compared to the standard Xbox Wireless Controller and even has significantly more features than Xbox Elite Series 2 at a much lower price. While it seems likely that it will rank among the best Xbox controllers on the market, there is one big catch to bear in mind.

The Scuf Valor Pro can only be used in a wired capacity and comes with a 10ft braided USB-C cable for that purpose. This might not be a huge deal for some, but it could make it difficult to fit the controller into your setup - especially if you play on the couch a fair distance from the TV.

In spite of this, the Scuf Valor Pro seems like a very compelling offering and it's great to see a traditionally pricey brand like Scuf branch out into lower cost products. The Scuf Valor Pro is available to order now via the Scuf website.

You might also like...

• Scuf Nomad review: a solid mobile controller that struggles to stand out
• Razer Wolverine V3 Pro review: a must-have premium Xbox controller
• The Mobile Controller MG-X Pro is the most disappointing mobile controller that I’ve ever tested

• The Sticky has been canceled by Prime Video after one season.
• The crime comedy series tells the true story of a maple syrup heist in Canada.
• Low viewership has been cited as the reason for its cancelation.

Prime Video seems to be in another sticky situation when it comes to canceling much-loved shows, as it’s axed crime comedy series The Sticky after one season.

The Sticky, which TechRadar's Lucy Buglass described as "Breaking Bad with maple syrup instead of meth" is based on a real-life maple syrup heist that happened in Canada in 2011. But despite its unique premise and positive reviews, the series failed to attract a wide enough audience according to Deadline.

As is generally the case with the best streaming services, Prime Video did not give an official reason for why the show had been scrapped, but it was likely due to poor viewing figures. The decision will come as a disappointment to fans of the show, especially as the season one finale set up a second season.

The Sticky failed to stick

The Sticky is loosely based on the real-life story of Canadian maple syrup farmer Ruth Landry (Margo Martindale), who turns to crime when greedy competitors threaten to shut down her business and take away her livelihood. In desperation, she teams up with a Boston gangster (Chris Diamantopoulos) and a French-Canadian security guard (Guillaume Cyr) to steal nearly 3,000 tonnes of maple syrup from a storage facility in Quebec.

The cancelation of one of the best Prime Video shows comes just two months since the first season, which guest starred Jamie Lee Curtis, was released on December 6, 2024.

The Sticky isn't the only comedy series to end up on Prime Video's scrap heap though, as it recently canceled The Pradeeps of Pittsburgh which also failed to secure a big viewership.

Comedies usually take longer to gain traction compared to drama series, which may explain why streamers are doing fewer comedy shows and why they generally have a higher cancelation rate. While Prime Video has produced comedy hits like Fleabag and The Marvelous Mrs. Maisel, it's failed to have a hit on the scale of award-winning Only Murders in the Building on Hulu and Disney+.

If you're wanting to see more heist action, the story that inspired The Sticky is also told in the Netflix documentary series Dirty Money, which has 100% on Rotten Tomatoes.

You might also like

• Amazon MGM Studios acquires the license to thrill as its gains full creative control of the entire James Bond franchise in landmark deal
• Lee Child reveals finding an actor for the big bad in Reacher season 3 that made Alan Ritchson look small was 'literally an international search'
• Invincible season 3 just included a sweet scene between Rex and Rae that's not in the comics, and now I'm more worried than ever for their safety

• Countries that tax US tech companies could face Trump tariffs
• DSTs come under fire by Trump in new memorandum
• Only the US can tax US tech companies, Trump argues

US President Trump has alluded countries that try to tax US based tech companies could face extra tariffs.

In the run up to the recent US election, Meta CEO Mark Zuckerberg called on Trump to stop the European Union from levying fines on US companies for violating data privacy and antitrust regulations.

In a recent memorandum, the President stated, “My Administration will not allow American companies and workers and American economic and national security interests to be compromised by one-sided, anti-competitive policies and practices of foreign governments.”

Tax or tariffs

Digital services taxes (DSTs) are one of the targets of Trump’s memorandum, of which the President has long been critical. DSTs were introduced to prevent tech companies from making profits in one country and collecting said profits in another country with more lenient taxation.

As The Register states, Netflix subscriptions were paid to a Netflix entity registered in the Netherlands, despite the users paying for their subscriptions from all over the world, thus dodging the taxation that would usually be paid for the economic activity generated in the user’s country.

“American businesses will no longer prop up failed foreign economies through extortive fines and taxes. All of these measures violate American sovereignty and offshore American jobs, limit American companies’ global competitiveness, and increase American operational costs while exposing our sensitive information to potentially hostile foreign regulators,” Trump’s memorandum states.

The Biden administration also was not a fan of DSTs, with the UK, Europe, and India dropping some of the DST regulations rather than facing 25% tariffs. DSTs were originally implemented as a temporary stop-gap alternative to the Organisation for Economic Co-operation and Development’s (OECD) wider plan to reduce tax-avoidance loopholes.

However, as many top US tech companies continue to avoid paying any taxes at all despite generating record profits, the OECD has been largely ineffective in its purpose, and any country attempting to levy taxes against US tech companies will face Trump’s wrath.

You might also like

• These are the best business VPN
• Take a look at our guide to the best password manager for families
• US lawmakers want Trump to call out UK Apple iCloud encryption backdoor demand

• A leaked CAD image suggests the iPhone 17 lineup could receive major design changes
• The iPhone 17 Pro and Pro Max might get a new, longer camera bar and a two-tone rear panel
• The iPhone 17 Air could also get a camera bar for its single camera

Though once known for pushing the boundaries of tech design, in recent years Apple has played it safe with even the best iPhones, generally leaning into the trend of homogenous, neutral slab phones that some find pleasingly minimal and others find straight-up boring (as a recent TechRadar poll confirms).

However, if a new leak is to be believed, this could all be set to change, with the iPhone 17 Pro and iPhone 17 Pro Max tipped to be the subjects of some big – and most likely controversial – design changes.

Noted leaker Majin Bu has shared what appears to be a leaked CAD mockup of the iPhone 17, iPhone 17 Air, iPhone 17 Pro, and iPhone 17 Pro Max on X (formerly Twitter), with the caption simply reading “iPhone 17 lineup CAD”.

(Image credit: Majin Bu )

CAD stands for Computer-aided design, and documents in this format are used to dictate the form and measurements of a product for automated assembly lines.

Though the base iPhone 17 model seems near enough unchanged from the current-generation iPhone 16, the other three models on show are either new or show off some serious changes.

Firstly, the phone to the far left of the image seems to be the all-new and much-rumored iPhone 17 Air. The image shows the new phone as sporting a single-lens camera system mounted on a wide pixel-style bar that protrudes from its slimmed-down chassis.

We’d previously heard rumors that the iPhone 17 Air would feature a camera bar cast in black to contrast against the rest of the unit, so this latest leak fits with our current expectations.

As for the iPhone 17 Pro and iPhone 17 Pro Max, the leak suggests there could be three main changes coming to the design of Apple’s highest-end handsets.

The first two concern the camera system: by the looks of things, the iPhone 17 Pro and iPhone 17 Pro Max feature a camera bar similar to the iPhone 17 Air, though obviously enlarged to provide space for the iconic triple-camera system. What appear to be the lidar sensor and flash module have both been moved to the far right side of this elongated camera housing.

An image supposedly showing the iPhone 17 Pro and a new clear case shared to X (formerly Twitter) by tipster Sonny Dickson also features this new camera design, with cutouts for the lidar sensor and flash on the right-hand side.

(Image credit: Sonny Dickson)

The third change is the one I’m betting will have the most Apple fans up in arms – the CAD render shows a separate cutout in the center of the iPhone 17 Pro and Pro Max’s rear panel, which could suggest a two-tone colorway or use of another material.

Given that even small changes to iPhone design – like the Pros switching to titanium rails or changing the ringer switch for the Action Button – can cause a storm of debate, it seems obvious that a two-tone iPhone would cause an explosion of Apple discourse.

Personally, I like it when phone makers get weird, and there’s no easier way to do this than by shaking up the design – though I can see why people might want to see Apple release something more subtle after several years of smaller iterations.

Anyhow, we don’t expect to hear anything official about the iPhone 17 lineup until much later in the year – as it stands, the above is based entirely on rumors. What do you think, though? Is the iPhone family due for a design shake-up? Or should Apple keep playing it safe? Let us know in the comments below.

You might also like

• More details of Samsung's tri-fold phone have leaked – and it could launch alongside the Galaxy Z Fold 7
• Leaked hands-on Samsung Galaxy S25 Edge video hints at its design and specs – and then disappears
• The Oppo Find N5 has made me even more excited for the Samsung Galaxy S25 Edge – here’s why

• A former Nintendo employee has discussed apparent Nintendo Switch 2 price leaks
• He said that retailers likely know "nothing" about the product's price or release date
• He added that retailers will probably learn the details around the same time that we do

Leak loving gamers beware, it looks like we shouldn't put much stock in any retailers when it comes to the Nintendo Switch 2's potential price or release date.

Speaking in a recent episode of the Kit & Krysta Podcast, a former Nintendo sales lead discussed the many online listings from various retailers that appear to give us an idea of how much the Nintendo Switch 2 could cost.

When asked how much retailers would know about the Nintendo Switch 2's price at this stage, he simply replied that they would know "nothing, certainly not price" and "almost certainly not any kind of launch date other than maybe a quarter."

He went on to describe retailers like Walmart Canada as among "the last people who are going to know anything about the price" and mentioned that, while some companies can speculate or guess, "they don't have that information."

"There's no way that even like Target in the US would know anything," he continued. "Nobody is going to know anything until we all know the launch date. When we all know the launch date, that's probably when the retailers are going to know the launch date [...] or maybe like a week or two in advance."

Speculation about the price of the Nintendo Switch 2 has been running rampant since it was first revealed. We're predicting that it will fall somewhere around $399.99 / £349.99 / AU$699.95. This would make it a little more expensive than the Nintendo Switch OLED, which launched at $349.99 / £309.99 / AU$539.95, to account for the upgraded specs, but not a figure that we feel would price out most of Nintendo's more casual audience.

In an investor Q&A, Nintendo also confirmed that the price of the Nintendo Switch 2 will be line with the "affordable prices customers expect" - which we suspect means a price point in line with our prediction. Of course, this is pure speculation right now. We'll likely have to wait for the Nintendo Switch 2 Direct on April 2 for official confirmation.

You might also like...

• 10 years later, Xenoblade Chronicles X Definitive Edition still has one of gaming’s best open worlds, but is it truly definitive?
• A Nintendo Switch 2 Pro Controller seems inevitable, so here are five things I want to see from a potential follow-up
• I’ve spent 20 hours rocking with the NEO S Purple Wave 5-Fret Fortnite Festival Edition and it’s a decent if nonessential way to play the rhythm game

• A new performance leak reveals that AMD's Radeon RX 9070 XT is 42% faster than the RX 7900 GRE at 4K
• Performance results could suggest a potential battle with the previous gen's flagship RX 7900 XT GPU, and Nvidia's RTX 5070 Ti
• AMD will reveal its new RDNA 4 GPU lineup on February 28

Nvidia has had its time to win fans over with the RTX 5000 series GPU launch, and things haven’t exactly gone smoothly for Team Green with GPUs shipping with missing ROPs (Render Output Units) and a host of other supply issues. AMD now has an opportunity to take advantage, and a new leak suggests the Radeon RX 9070 XT might do just that.

According to a leaked figure from AMD (pictured below, originally spotted by VideoCardz), the new Radeon RX 9070 XT runs 42% faster than the Radeon RX 7900 GRE at 4K ultra settings across multiple games, hinting at potential competition with Nvidia's RTX 5070 Ti. The performance gains over the previous generation's GPU are seen in both raster and ray tracing performance, so the full picture isn't clear just yet, but it’s certainly promising.

(Image credit: VideoCardz)

It's important to note that the RX 7900 GRE is the strongest point of comparison here - none of Nvidia's RTX 4000 or 5000 series GPUs are used for reference. The GRE was formerly a China-only GPU that was released globally following positive reception, though it’s less powerful than the RX 7900 XT, sitting between Nvidia’s RTX 4070 Super and 4070 Ti in terms of gaming performance. However, based on the suggested 42% performance difference, this could indicate the new RDNA 4 GPU's position against both the previous flagship RX 7900 XTX and Nvidia's RTX 5070 Ti.

Team Red will showcase its Radeon RX 9000 series GPU lineup on February 28 in preparation for an early March launch - here, we'll likely see the pricing of the cards and FSR 4's full unveiling, which already has some stiff competition in Nvidia's impressive DLSS 4.

(Image credit: VideoCardz) AMD, it's now or never...

Nvidia is currently facing some backlash over the launch issues surrounding its Blackwell GPUs (as well as the sky-high pricing on the high-end cards), so there's no better time than now for AMD to capitalize. Hardly any information is available regarding the new Radeon RX 9000 series GPUs - excluding its brief reveal at CES 2025 and a range of leaks and rumors - so there's plenty of room for surprises.

Team Red has promised upgrades to ray tracing with the RDNA 4 architecture, something its fierce rival Nvidia has essentially mastered at this point. The previous Radeon generation struggled with ray tracing, with the flagship RX 7900 XTX’s raster performance against other cards being the highlight instead, so it's promising to hear that there will be a greater focus on other improvements with the new Radeon GPU lineup.

To put it lightly, Nvidia is in a bit of a tight spot with a plethora of issues present across all three released GPUs - the RTX 5090, RTX 5070, and RTX 5070 Ti have all been affected by the missing ROPs issue, which won't sit well with consumers, especially given the high price of admission for those cards. This could give Team Red the momentum required to win fans over, with appealing GPU performance results at affordable prices.

It's clear that these new GPUs from AMD aren't supposed to compete at the high end (with the now-debunked 32GB GPU rumor put to bed). But if the RX 9070 XT can compete with Team Green's RTX 5070 Ti without stock issues and inflated prices, we could have a mid-range winner on our hands.

You may also like...

• AMD RX 9070 GPU spec and benchmark rumors cast fresh doubt on power usage and performance – but I wouldn’t worry about the latter
• Watch out, Nvidia – AMD’s rumored to have a plan to unleash RX 9070 reviews to run interference with RTX 5070 GPU’s big launch day
• AMD’s RX 9070 GPUs could go on sale March 6, the day after Nvidia’s RTX 5070 – and I wouldn’t fret about those 900W power supply rumors

• Sigma's minimalist BF is machined from a single piece of aluminum
• It features Sigma's L-mount and a 24MP full-frame sensor with 6K video
• There's only 230GB of internal storage, and no memory card slot

Sigma has clearly had fun designing its new BF full-frame mirrorless camera, stripping the user experience right back to the basics. It's made in Japan, but adopts Scandinavian minimalist design and then ups the ante.

As Sigma writes, the BF is a "radically simple camera that evokes the essence of photography". This is no boring black box with full-frame sensor like we tend to get from Sony, Canon, Nikon and Panasonic – it is indeed something radically different, and I respect Sigma for its boldness, even if the BF might be easy on the eye over practical in the hand.

The BF should be as tough as they come – it's machined from a single piece of aluminum, a process which Sigma says takes over seven hours. It measures just 5.1 x 2.9 x 1.4in / 130.1 x 72.8 x 36.8mm and tips the scales at 15.7oz / 446g (with battery), and accepts L-mount lenses, of which there are many, making it suitable as an everyday camera.

There are two colors to choose between: silver or black. Alongside the camera, Sigma unveiled silver versions of its I-series of contemporary lenses to match the look of the silver option. The camera's list price is $1,999 / £1,969, which is pretty competitive considering the materials and craftsmanship involved. We'll update this article with Australia pricing when we get that info.

Within its sleek exterior, the Sigma BF houses a full-frame 24MP sensor with 6K video capabilities and an electronic shutter only – there's no mechanical shutter. A scan of the camera's features suggests that the Sigma BF is pretty similar to the Panasonic Lumix S9, which is another compact full-frame mirrorless camera.

I tested the Lumix S9, and I felt that Panasonic had made one compromise too many in the service of keeping things simple for me to truly enjoy its handling (even if a firmware update improved matters somewhat), and Sigma's BF is intentionally even simpler. I've not had my hands on the camera yet, but I have developed a few thoughts already.

(Image credit: Sigma) It's bold, exciting – and limited Two Sigma lenses also added to the collection

Alongside the BF camera, Sigma unveiled two lenses; the 300-600mm F4 DG OS for full-frame cameras, and the 16-300mm F3.5-6.7 DC OS for APS-C cameras. The 300-600mm professional lens costs $5,999 / £5,899, while the 16-300mm is the world's first mirrorless lens with 18.8x optical zoom, and costs $699 / £599. Both lenses ship in April. Again, Australia pricing is TBC.

I fully respect that in order to create a radically simple camera, some features have to go. I'm also writing as a seasoned photographer who's used to enjoying the number and variety of controls that complex cameras offer, even if a personal favorite is the wonderfully simple Hasselblad X2D 100C. But I can already sense that the Sigma BF is a camera that I'll enjoy looking at more than shooting with.

For me, the challenge with using the BF will be composing images. There's no viewfinder, which is a big minus for me given that the BF is a photography-first camera. This is compounded by the 3.15-inch 2.1M dot touchscreen, which is fixed – you can't flip it out or spin it around for easy viewing at awkward angles.

Such limitations are the price we pay for a radically minimalist camera. Still, I'm intrigued by the Sigma BF, and it could well be one of 2025's standout cameras. I love how its menus and controls are stripped back to what you need the most – there are just four buttons to access basic exposure settings and playback, all controlled just with one finger, together with a small rear LCD display which displays your most recent changed setting. Check out Sigma's Youtube video below for a closer look at how the controls work.

Around the sides and underside, there's just single loop for a wrist strap, a tripod mount, a removable battery and a USB-C port; no other ports for headphones and mic inputs. This is as simple as it comes. In fact, there isn't even a memory card slot – instead there's 230GB of internal storage only.

I don't much mind of any of those omissions, and nor do I mind that sleek front with no grip. After all, this is a camera that's designed for small lenses and for everyday photography.

For me, it's the lack of viewfinder and fixed rear screen that are dealbreakers. What do you think of the new Sigma BF? Let me know in the comments below.

You might also like

• I’ve tested all 4 of Sigma’s bargain f/1.4 primes for Canon cameras – here’s how to decide which one to buy
• Leica could launch a controversial M-series camera this year with a new kind of viewfinder – it sounds like function over soul
• Best full-frame mirrorless camera 2025: top options from Canon, Sony, Nikon and more

DALL-E 3 is one of the most advanced AI image generators, capable of creating realistic images based on simple text prompts.

The text-to-image tool was developed by OpenAI, creators of ChatGPT. Simple yet powerful, DALL-E 3 can process detailed written descriptions and generate matching images in a range of styles, including illustrations.

Trained on millions of existing images, DALL-E 3 is improving all the time. Read on to find out how you can use it.

This article was correct as of February 2025. AI tools are updated regularly and it is possible that some features have changed since this article was written. Some features may also only be available in certain countries.

What is DALL-E 3?

DALL-E launched in January 2021. It was the first AI-powered text-to-image tool to gain widespread attention. Developed by OpenAI, the model uses deep learning to convert written descriptions into visual assets.

Announced in September 2023, DALL-E 3 is the most advanced version to date. Trained on a huge dataset of images and text descriptions, it’s capable of understanding complex natural language prompts with greater accuracy than previous generations, allowing even complete beginners to produce vivid, detailed images using the tool.

Integrated into ChatGPT, DALL-E 3 is able to generate images at higher resolutions than before. It can also understand and replicate a range of aesthetic styles. Users can selectively edit and refine images by fine-tuning their text instructions.

What can you use DALL-E 3 for?

DALL-E 3 makes it easy to generate images based on text-based descriptions written in natural language. It’s a useful tool for creatives to quickly explore and experiment with visual concepts. It also allows users without design skills to easily create graphics.

Image prompts can be as straightforward or as complex as you like. DALL-E 3 is capable of following details with a good degree of accuracy. DALL-E 3 can also create images in a range of styles, including photorealistic, illustrated, watercolor and line art.

ChatGPT suggests using the following basic structure for image prompts with DALL-E 3: subject + environment/background + style + mood/lighting + specific details.

The resulting images can be used for everything from movie posters and brand logos to tattoo designs and printable graphic artwork.

What can’t you use DALL-E 3 for?

DALL-E 3 has a number of content limitations, most of which are for legal reasons. It can’t be used to generate images of copyrighted characters, content or logos. It also can’t be used to replicate the signature style of a living artist.

Naturally, DALL-E 3 prohibits the creation of inappropriate content, including explicit or violent imagery.

DALL-E 3 also can’t be used to produce misleading imagery, nor will it generate images of private individuals. If you ask it to create an image of a celebrity, it will offer an image “inspired by” their likeness.

Beyond these restrictions, DALL-E 3 can only produce still images, with no capacity to create videos. It can’t be used to generate very high resolution images, either. Image dimensions are limited to the following:

• 1024 × 1024 pixels (square)
• 1792 × 1024 pixels (portrait)
• 1024 × 1792 pixels (landscape)

DALL-E 3 also can’t be used to produce images in large batches. It can generate up to four images at a time.

How much does DALL-E 3 cost?

You can access DALL-E 3 for free through the Image Creator tool in Microsoft Designer, an online platform. You get 15 complimentary credits per month. Each image that you generate uses one credit.

For more credits, you’ll need to take out one of the following plans:

• Microsoft 365 Personal: $9.99 / £8.49 / AU$16 per month or $99.99 / £84.99 / AU$159 per year
• Microsoft365 Family: $12.99 / £10.49 / AU$18 per month or $129.99 / £104.99 / AU$179 per year
• Copilot Pro: $20 / £19 / AU$33 per month

Alternatively, you can use DALL-E 3 through ChatGPT with a ChatGPT Plus subscription, which costs $20 / £19 /AU$33 a month. You can also use it with a free ChatGPT account, although the number of images you can create per day is limited to about 3.

Where can you use DALL-E 3?

There are two ways to use DALL-E 3. The first is through ChatGPT. You can access it on the web, via the desktop app for macOS and Windows, or through the smartphone app for Android and iOS. With a ChatGPT Plus subscription, simply enter your request like any other text prompt in the dialogue box and ChatGPT will generate an image using DALL-E 3, or you can select DALL-E in the menu at the top left of the web interface.

Alternatively, you can use DALL-E 3 through the Image Creator tool in Microsoft Designer. This is a web-based tool. Enter your text-based description, select the desired output size and hit generate.

Is DALL-E 3 any good?

DALL-E 3 is one of the most user-friendly AI text-to-image generators you'll find. Creating and editing images is as simple as entering text prompts through its intuitive web-based interface.

We compared DALL-E 3 to Midjourney on TechRadar and while our writer praised the vivid quality of images generated by DALL-E 3, an aesthetic which lends itself well to visual storytelling, he didn't think it was as good as Midjourney at creating realistic images.

With DALL-E 3 you also get just one image per prompt, while Midjourney gives you four to choose from. That said, DALL-E 3 is a highly commended AI image generator and if you're already paying for a ChatGPT membership it comes as part of the package.

Use DALL-E 3 if...

You’re looking for an intuitive tool

Creating images with DALL-E 3 is as simple as typing word prompts into a text box, using either ChatGPT Plus or Microsoft’s free Image Creator. You can choose different output sizes and selectively edit the results on both platforms.

You need accurate text in your images

DALL-E 3 is effective at processing complex natural language prompts. It’s notably better than other AI image generators at rendering accurate text in images, following instructions to the letter.

Don't use DALL-E 3 if...

You need photorealistic image assets

DALL-E 3 can produce vivid and dynamic images, but photorealism isn’t its strong suit: results tend to have a noticeably synthetic look. This aesthetic makes the tool more effective when tasked with creating stylized graphics, cartoons or artistic images.

You want an unlimited image editor

You can create a limited number of free images with DALL-E 3 using Microsoft Designer’s Image Creator tool. When your free credits run out, you’ll need to take out a paid plan with Microsoft or ChatGPT to access it.

Also consider

• Midjourney is an AI text-to-image generator accessed via a dedicated Discord chat channel. Its user interface isn’t the slickest, but the platform has useful editing tools and is capable of creating distinctive images with artistic lighting and detail.
• Leonardo is a well-equipped image generator for professionals, with useful features such as real-time editing, upscaling and the option to train models on brand assets. Its UI can be clunky, but it still offers great value for money.

You might also like...

• Hands-on with the best AI image generators
• Leonardo vs. Midjourney: AI image generators put to the test
• Dall-E vs. Midjourney: which is the best AI image generator?

• Users report that one of their earbuds just won't charge
• Lots of suggested solutions don't seem to work
• Some users reported burn marks near charging pins

It looks like the Samsung Galaxy Buds 3 Pro aren't out of the quality control woods just yet. Last summer the buds were delayed due to quality control issues of multiple kinds; now, it seems there are still serious issues with at least some users' earbuds.

According to Sammyfans, multiple Buds 3 Pro users are reporting that one of their earbuds simply won't charge.

While several users have suggested things that may have helped them to get things started, there doesn't seem to be any trick that's working more broadly.

(Image credit: Samsung) What's the charging problem with Galaxy Buds 3 Pro?

According to the report, users have tried restarting their phones, resetting the earbuds to factory settings, checking for debris and, of course, checked the really basic stuff such as ensuring the charging case is charged. But it seems that these attempts haven't been successful in finding a consistent fix.

There's one bit in the report that's particularly worrying. "Some users have even noted burn marks near the charging pins," which is not something you want to see in your expensive electronics, though obviously we'll have to take it with a grain of salt at this time as to whether this is totally accurate, or related.

The strongest culprit so far is that the issue may be due to imperfectly positioned charging pins: some users have reported successfully charging their errant earbuds by holding them carefully in position rather than just sitting them in the case, or even by gently adjusting the charging pin position with a small screwdriver.

If you're affected by the charging issue, I wouldn't recommend getting the screwdrivers out, or at least I wouldn't if your DIY skills are as destructive as mine: if this is a fault, it should be a warranty fix under Samsung's standard 12-month warranty, so your first port of call should be Samsung support.

We rate the Samsung Buds3 Pro as being among the best wireless earbuds for those with Samsung phones, but we'll be keeping an eye on this story to see if it develops.

You might also like

• The best earbuds for every budget
• On a budget? The best cheap wireless earbuds are here
• Want noise-cancelling earbuds ready for 2025 travel? Here are the 4 best we’ve tested for all budgets

• Norton is adding AI-powered scam protection directly to its plans
• The tools cover SMS messages, emails, web browsing, and more
• It comes free for all with purchased Norton plans

Norton is adding new AI tools to its solutions, to help its customers better protect from internet scams.

The antivirus giant has announced Genie Scam Protection and Genie Scam Protection Pro, which it says provide “advanced AI protection in all the usual hotbeds of scams: texts, phone calls, emails, and web.”

Norton says that Genie AI analyzes “the meaning of words, not just links” to spot hidden scam patterns that even the most trained eye can miss. Norton’s most comprehensive plan, Norton 360 with LifeLock Ultimate Plus, also offers Scam support and Reimbursement coverage.

Fighting AI with AI

The company introduced Norton Genie back in 2023, to help users analyze suspicious messages and images in real-time. Now, by building it directly into Norton plans, the company wants to build a “set it and forget it” solution that works proactively to keep people safer.

Whoever bought Norton’s Cyber Safety products on desktop, or mobile, (including Norton AntiVirus Plus, Norton Mobile Security, and Norton 360), will get the Genie Scam Protection for free.

Depending on the set of purchased features, users can expect the AI to help scan SMS messages, internet websites, calls, emails, and more. They will also get an AI-powered “scam assistant”, which integrates the Norton Genie AI app to provide “instant guidance on scams and suspicious offers”.

The tools are available in the US now, with other regions soon to follow.

For Leena Elias, Chief Product Officer at Gen, adding AI to cybersecurity solutions is a logical step, since crooks have been using AI in their attacks for a while now.

“Scammers are tapping into AI to speed up their schemes and make them more believable,” Elias said. “Couple that with the sheer volume of emails, texts, and calls we’re getting every day; it’s clear we need technology on our side to help us avoid becoming the next scam victim.”

In the near future, Norton says it will also integrate deepfake and scam detection into AI-supported PCs.

You might also like

• These are the most common PayPal scams around right now - so stay alert
• We've rounded up the best password managers
• Take a look at our guide to the best authenticator app

• Apple has released its iOS 18.4 beta
• Code indicates Google Gemini could soon be integrated into Siri
• It would allow users to ask Google AI more complicated queries

Gemini could be integrated into Siri very soon, as new code in the iOS 18.4 beta hints at more AI model functionality for Apple Intelligence.

Google Gemini functionality in Apple Intelligence-powered Siri would allow users to ask Google AI more difficult questions than Siri can currently answer. This would be available as an alternate option to ChatGPT integration, although we’re not sure if you’ll need to choose or if you’ll be able to access different AI models depending on your queries.

At WWDC 2024 when Apple unveiled Apple Intelligence, Craig Federighi said, “We also intend to add support for other AI models in the future.” following the ChatGPT Siri integration announcement, it would now appear that Gemini is the next AI model to follow suit.

According to @aarop613 on X, there is a reference to “Google” as part of “choice_options” alongside the already available ChatGPT integration. This would imply that Gemini might be available in Siri as soon as the official iOS 18.4 launch which is expected sometime in the next few months.

Google Gemini coming to iPhone?

Apple's backend has revealed that the next 3rd party iOS-AI integration will be Google Gemini pic.twitter.com/0rIuJhT5LjFebruary 21, 2025

This wouldn’t be the first time Apple and Google work together, the companies have had a lucrative deal to include Google Search as the default search engine in Safari on iOS for years now, and despite being direct competitors see benefits in working together.

Google recently released Gemini 2.0 Flash which is an incredible multimodal AI model that can go toe-to-toe with the best OpenAI has to offer. Implementing Gemini into Siri would give Apple Intelligence an edge over other smartphone AI offerings by incorporating multiple models into one user experience.

If Apple’s goal is to continuously increase AI model offerings in Siri with Apple Intelligence, users will benefit from a wide scope of functionality, and that’s only a good thing. While we don’t know when to expect Gemini integration in Siri, this code points towards the arrival of Google AI in the not-too-distant future.

You may also like

• Free Gemini Live update brings better conversation skills and understanding of accents
• You only need to see this screenshot once to realize why Gemini needs to follow ChatGPT in making its AI products less confusing
• I installed iOS 18.4 dev beta and the big Siri intelligence update is nowhere to be found

• A timer has appeared on the official Tony Hawk game website
• This follows an easter egg that was spotted in Black Ops 6 last week
• Developer Iron Galaxy seems to be involved

Alert: the ongoing Tony Hawk's Pro Skater mystery has intensified as a timer has appeared on the series' official website.

By heading over to the website, you'll find a date, March 4, 2025, as well as a timer that (at the time of writing) has roughly eight and a half days left to go. This lines up with an easter egg from Call of Duty: Black Ops 6 which was spotted last week in the Grind multiplayer map, in which a sign showing off that same date could be found in the skate park-themed environment.

It certainly seems like something is in the works, then, but whether that's a brand new game or the highly requested Tony Hawk's Pro Skater 3+4 remaster remains to be seen.

What we can glean from the website is the ESRB Teen rating that highlights 'Lyrics' as a content warning. Par for the course for the THPS series, really, given its usage of licensed music. What's more interesting is the listing of developer Iron Galaxy next to the Activision logo.

Iron Galaxy is a pretty renowned developer in the industry. It has worked with Activision in the past, having ported Crash Bandicoot N. Sane Trilogy, Spyro Reignited Trilogy and Tony Hawk's Pro Skater 1+2 to PC. The dev also has a pretty respectable background in fighting games, having worked on Street Fighter 3: Third Strike Online Edition, Killer Instinct (2013) as well as the unfortunately discontinued Rumbleverse.

So what will Activision have in store for us on March 4? We simply won't know until then, so in the meantime we'll just have to blast some Reel Big Fish, wear our favorite pair of checkerboard Vans slip-ons and hope for the best.

You might also like...

• Why are we getting so many PlayStation PC ports? Former president says it's 'like printing money'
• 10 years later, Xenoblade Chronicles X Definitive Edition still has one of gaming’s best open worlds, but is it truly definitive?
• Elden Ring Nightreign has the makings of a seriously fun multiplayer game if FromSoftware can fix its performance issues

• A new report about the Samsung tri-fold phone has revealed more details
• It may appear alongside Samsung's other 2025 foldables
• A Galaxy G Fold name has already been rumored

Last month, Samsung teased a tri-fold foldable phone that's set to launch sometime later this year – and we've now got more leaked information about what to expect when the new device eventually shows up.

As per South Korean outlet ET News (via @Jukanlosreve), mass production of the foldable is scheduled for April, as Samsung finalizes supply chain parts. Unlike the Huawei Mate XT, it will apparently fold inwards (as previously rumored), so two screens are needed.

Those screens will measure, according to this report, 9.96 inches and 6.49 inches. That main screen is just a touch smaller than the Huawei tri-fold, which has a 10.2-inch main display (a third of which does double duty as a cover display).

Another important revelation here: the tri-fold is being tipped to launch alongside the Samsung Galaxy Z Fold 7, the Samsung Galaxy Z Flip 7, and the Samsung Galaxy Z Fold SE 2, which should be around July time – though it won't necessarily go on sale right away.

A rumored name

Report: Samsung Prepares for Mass Production of ‘Dual-Folding’ Foldable Phone – Expected to Launch Alongside Z7 SeriesAccording to reports from Korean media citing anonymous sources, Samsung plans to start procuring components for its dual-folding foldable phone around April.…February 24, 2025

This ET News report also suggests the main screen will lack a digitizer, in order to keep it as slim as possible – that means no support for an S Pen stylus, unlike the Samsung Galaxy Z Fold 6 and the Samsung Galaxy S25 Ultra.

We don't get a name for the tri-fold device here, but an earlier rumor indicated it's going to be called the Samsung G Fold, which would sort of fit in with the current naming scheme for Samsung's foldable series.

(Image credit: Future)

The upcoming handset is apparently going to match the current Galaxy Z Fold 6 in terms of screen brightness (2,600 nits) and wired charging speed (25W), but we're still mostly in the dark when it comes to the other key specs.

It certainly sounds as though we're going to get a bumper crop of foldables from Samsung in the next few months, but the Galaxy G Fold (if that is what it's called) might be the best of the bunch – even if it's also very expensive.

You might also like

• Samsung maybe just solved the tri-fold naming problem
• These are the best foldables you can buy
• Samsung shows off two tri-fold concept phones at CES 2025

• The M4 MacBook Air release date could be imminent
• Apple Stores are getting ready and cutting stock of the M3 MacBook Air
• The new laptop is expected to bring modest improvements

We’ve been hearing for months that Apple is going to launch a new MacBook Air with an M4 chip soon, and now it looks like the date is almost upon us. If you’ve been thinking about buying a MacBook Air, you’ll want to hold off for just a little longer.

That’s because Bloomberg reporter Mark Gurman has claimed in his Power On newsletter that Apple Stores are preparing for the new MacBook Air to arrive. His evidence for this? Inventory of the M3 MacBook Air is apparently being wound down at Apple Stores, meaning there will soon be fewer units of the current model available to buy.

This is usually a surefire sign that Apple is getting ready to launch a new device. After all, what’s the point in the company maintaining high stock levels of a product that will soon be superseded?

It’s not the only clue. Gurman says that Apple’s marketing, sales and retail teams are now being prepared for the M4 MacBook Air release, which is expected to land at some point in March. Gurman's track record is very good, so a launch within the next month does look increasingly likely.

Hang on a little longer

(Image credit: Apple)

All of this means that you probably shouldn’t buy a MacBook Air right now. If you do, you’ll end up with an M3 model that will quickly be surpassed by the M4 version.

As we’ve seen from the MacBook Pro and the iPad Pro, the M4 chip should be a decent upgrade over the M3, if not a groundbreaking one. In terms of performance, we’ll likely see some increased core counts, a more efficient Neural Engine, and improved memory bandwidth.

Other than the new chip, the M4 MacBook Air isn’t expected to include many other new features. The design will almost certainly stay the same, as will the speakers. And while Apple is working on an OLED MacBook Air, it’s still a few years away from release.

Still, the M4 MacBook Air is enough of an upgrade to make it worth waiting for. If you’ve been itching to pull the trigger on a new MacBook Air, you’d do well to hang on just a few more weeks.

You might also like

• Apple MacBook Air M4: news, rumors, and everything we know
• MacBook Air OLED reportedly delayed until at least 2028 – here’s why
• Apple M4 chip: everything you need to know about Apple's latest silicon

• JBL Flip 7 and Charge 6 US launch details leak in removed press release
• Pre-orders set for March 10, 2025 – on-sale April 6, 2025
• Flip 6 set to be $149, Charge 6 set to be $199

The JBL Flip 7 and Charge 6 Bluetooth speakers are ready to launch soon, according to a press release from the company on Business Wire, which has now been removed – but is still visible via Wayback Machine.

It reveals that the new speakers will cost $149 and $199 respectively, and will be available to pre-order in the US from March 10, 2025 – but won't actually be available until April 6. The price is a rise from the previous versions of both models, which launched at $129 and $179 respectively. In the UK, we'd expect them to cost £149 and £199 based on current exchange rates. In Australia, it works out to around AU$295 and AU$395.

That's unless you're in China, where the Flip 7 it's already available to order with an early bird price of around $158, according to Gizmochina.

In addition to the usual JBL strengths, it looks like the successors to the rather good JBL Flip 6 and JBL Charge 5 will be borrowing one of the best features of the JBL Xtreme 4: the AI Sound Boost mode. The Business Wire release confirms that both speakers will support this tech, and we'll explain why that's a great fit in a moment.

But the Gizmochina article says that there's another welcome piece of optimization tech in the Flip 7: a battery mode that squeezes another two hours out of the high-capacity lithium-ion battery by turning off the bass display and equalizer settings. That should mean an improved battery life of 14 hours.

We haven't seen a good range of pictures of the devices yet, but we expect the design to be largely to same as the previous models – here's the JBL Charge 5, for reference. (Image credit: Daniel Hessel) Why we're looking forward to hearing the JBL Flip 7

I'm the first person to mock fancy-sounding audio optimization tech, because I've listened to a lot of it and found it deeply disappointing. But the AI Sound Boost here could be good, based on when we've already heard it in action.

Here's Harry Padoan, who reviewed the JBL Xtreme 4, describing how the system worked for him: "The AI Sound Boost in the JBL Xtreme 4 has pretty controlled, powerful audio and at high volumes it didn't sound distorted at all – that seems to be the purpose of the technology. How much of that is really down to 'AI', I couldn't say, of course."

Whether it's AI, engineering or a mix of both doesn't really matter: if it helps the JBL Flip 7 and Charge to push out more powerful sound without distortion than they could manage before, that could be a big win for such small speakers.

The Gizmochina articles describes the Flip 7 as using a tweeter, racetrack mid driver and dual bass radiators for sound, which is basically what the Flip 6 offers.

It'll be IP68 water and dust-resistant, but a nice upgrade will be the option of USB-C audio input for "lossless" sound, though I don't know that I'd rely on a pint-sized portable speaker for that kind of fidelity…

Bluetooth 5.4 will provide the wireless tech, including support for Auracast, to provide the ability to use loads of the speakers together at once.

We don't have this kind of info for the Charge 6 yet, but I'd expect them to be basically identical except for a more powerful speaker setup in the Charge, and a larger battery that can charge other devices, naturally.

The JBL Flip 6 and JBL Charge 5 both rank among our list of the best Bluetooth speakers based on how good a value they offer, and I suspect that these two speakers will be the biggest competitors to the new models – with the Flip 6 regularly available for under $100 / £100, the Flip 7 is going to have to be a clear upgrade.

You might also like

• Acer warns of tariff-related price hikes for tech products
• The best Bluetooth speakers
• The best wireless speakers including Wi-Fi models

• Windows 11 has a new preview build with some smart additions
• One of those is a fresh piece of the interface to make file sharing easier
• The other is a couple of neat new layouts for one part of the Start menu

Windows 11 should be getting some useful changes to its interface, one of which makes sharing files easier, and the other introduces nifty new layout options for the Start menu.

These moves are in testing right now, to be precise in Windows 11’s build 22635.4950 which is in the Beta channel.

The new file sharing menu appears at the top of the screen whenever you’re dragging a file out of a folder on the desktop, and it features an array of icons, such as WhatsApp, My Phone, Microsoft Teams, and so on. Dropping a file on any icon will instantly share it via that medium.

As for the Start menu, that change is for the ‘All apps’ page which as the name suggests lists all your apps, with the new options here being the grid and category views.

The grid view sticks to an alphabetical ordering of your apps, as in the standard list view, except with the grid layout, you can fit more apps on with the layout extended out to the side, so there’s less scrolling to do (see the image below).

(Image credit: Microsoft)

Or a more different alternative is the category view where your apps are grouped into various categories and ordered by usage levels. So, for example, there are social apps, or productivity, or entertainment apps, a whole bunch of categories you can dive into, with your most-used apps highlighted in a box featuring their icons (again, see the pic below).

(Image credit: Microsoft)

This feature is still rolling out, so not every tester will see it yet (and it’s coming to the Dev builds, as well as the Beta channel).

That’s pretty much it for this preview build, although there’s one more addition on the accessibility front with File Explorer which is being graced with improved support for text size scaling with File Open or Save dialog boxes (and also with Wizard dialog boxes).

Oh, and Microsoft has applied a number of bug fixes too, while there are some acknowledged issues, which are all listed in the usual blog post outlining the content of build 22635.4950.

Analysis: Almost through testing now

Of course, all this is just in testing, and if trialling any given element doesn’t work out, that feature may be dropped. I’d imagine that’s unlikely, though, as both these fresh introductions are useful and welcome touches.

Easier file sharing has got to be good, and I can’t see anyone arguing with more choice in the Start menu layouts. After all, if you don’t want to have the ‘All apps’ panel look any different to the default list, you don’t need to go near the alternatives.

It’s also worth bearing in mind that having reached the Beta channel, these features are not far from release now. There’s only one more step – final testing in the Release Preview channel – before this functionality is ready to be deployed to the finished version of Windows 11.

Indeed, the move with the Start menu layouts was previously discovered by digging around in test builds in the past, so it’s good to see that coming to fruition (the above mentioned caveats taken into consideration).

You may also like...

• Windows 11 fully streamlined in just two clicks? Talon utility promises to rip all the bloatware out of Microsoft’s OS in a hassle-free way
• Are you unable to get security updates for Windows 11 24H2? Here’s the likely reason why, and the fix to get your PC safe and secure again
• If Microsoft's Windows 11 24H2 issues continue like this, I'll fully convert to SteamOS for gaming