SAVE 32%: As of July 6, you can get the Apple AirPods Pro 2 for $169.99, down from $249. That’s a 32% discount.
Amazon is rolling out its Prime Day deals early this year, and some of the discounts are seriously impressive. With the main event just two days away, we’re seeing killer deals across every category, and Apple devices (discounted MacBooks, AirTags, iPads, and more) are getting some of the best price cuts.
SEE ALSO:
Amazon Prime Day vs. Black Friday: When are the deals better?
As of July 6, you can get the Apple AirPods Pro 2 for $169.99, down from $249. That’s a 32% discount and a $79.01 savings. It’s not the absolute lowest price we’ve ever seen (they hit $153.99 during Black Friday), but it’s still a solid deal.
The AirPods Pro 2 feature impressive sound control, with Active Noise Cancellation (ANC), a Transparency Mode, and an Adaptive Audio setting that automatically blends the two. The physical design is just as thoughtful: you get four sizes of silicone tips for a secure fit, and you can now swipe the stems to adjust the volume. They’re also built for everyday use, with an IP54 rating against sweat and rain, and a battery that provides up to 30 hours of total listening time.
Mashable Deals
SEE ALSO:
The new AirPods Pro look old but sound fresh
They also have a powerful, built-in Hearing Aid feature. It’s a clinical-grade function for users with perceived mild to moderate hearing loss that works by having you take a hearing test on your iPhone to create a custom sound profile. This also enables Media Assist, a feature that uses your profile to automatically make music, videos, and phone calls sound sharper and easier to understand.
The best early Prime Day deals, hand-picked by Mashable’s team of experts
In a recent paper, Stanford researchers Mason Kamb and Surya Ganguli proposed a mechanism that could underlie the creativity of diffusion models. The mathematical model they developed suggests that this creativity is a deterministic consequence of how those models use the denoising process to generate images.
In rough terms, diffusion models are trained to sort of uncover an image from an isotropic Gaussian noise distribution that is the outcome of the training process from a finite set of sample images. This process consists of gradually removing the Gaussian noise by learning a scoring function that points in gradient directions of increasing probability.
If the network can learn this ideal score function exactly, then they will implement a perfect reversal of the forward process. This, in turn, will only be able to turn Gaussian noise into memorized training examples.
This means that, to generate new images that are far from the training set, the models must fail to learn the ideal score (IS) function. One way to explain how this occurs is by hypothesizing the presence of inductive biases that may provide a more exact account of what diffusion models are actually doing when creatively generating new samples.
By analyzing how diffusion models estimate the score function using CNNs, the researchers identify two such biases: translational equivariance and locality. Translational equivariance refers to the model’s tendency to reflect shifts in the input image, meaning that if the input is shifted by a few pixels, the generated image will mirror that shift. Locality, on the other hand, arises from the convolutional neural networks (CNNs) used to learn the score function, which only consider a small neighborhood of input pixels rather than the entire image.
Based on these insights, the researchers built a mathematical model aimed at optimizing a score function for equivariance and locality, which they called an equivariant local score (ELS) machine.
An ELS machine is a set of equations that can calculate the composition of denoised images and compared its output with that of diffusion models such as ResNets and UNets trained on simplified models. What they found was “a remarkable and uniform quantitative agreement between the CNN outputs and ELS machine outputs”, with an accuracy of around 90% or higher depending on the acutal diffusion model and dataset considered.
To our knowledge, this is the first time an analytic theory has explained the creative outputs of a trained deep neural network-based generative model to this level of accuracy. Importantly, the (E)LS machine explains all trained outputs far better than the IS machine.
According to Ganguli, their research explains how diffusion model create new images “by mixing and matching different local training set image patches at different locations in the new output, yielding a local patch mosaic model of creativity”. The theory also helps explain why diffusion models make mistakes, for example generating excess fingers or limbs, due to excessive locality.
This result, while compelling, initially excluded diffusion models that incorporate highly non-local self-attention (SA) layers, which violate the locality assumption in the researchers’ hypothesis. To address this, the authors used their ELS machine to predict the output of a publicly available UNet+SA model pretrained on CIFAR-10 and found that it still achieved significantly higher accuracy than the baseline IS machine.
According to the researchers, their results suggest that locality and equivariance are sufficient to explain the creativity of convolution-only diffusion models and could form the foundation for further study of more complex diffusion models.
The researchers also shared the code they used to train the diffusion models they used in the study.
Do you need help? Please contact our support team from 9 to 17 CET via support.pressclub@bmwgroup.com.
PressClub Global · Article.
Sun Jul 06 18:01:20 CEST 2025Press Release
+++ Daniel Brown triumphs in Munich with a total score of 22 under par +++ Second place goes to Jordan Smith (ENG) ahead of Kazuma Kobori (NZL)+++ Matti Schmid (-15) best German player +++ Images from the 18th green +++
+++ Daniel Brown triumphs in Munich with a total score of 22 under
par +++ Second place goes to Jordan Smith (ENG) ahead of Kazuma Kobori
(NZL)+++ Matti Schmid (-15) best German player +++ Images from the
18th green +++
Munich. After a high-class and thrilling final round,
Daniel Brown secured the coveted title at the 36th BMW International
Open. The Englishman needed 66 strokes on Sunday on the Golfclub
München Eichenried Championship Course and triumphed with a total
score of 266 strokes (22 under par). As the best German player, Matti
Schmid (-15) finished the tournament tied for 7th place. Tim
Wiedemeyer (-12) was the best amateur, achieving an impressive tied
17th rank.
Below, you will find recent images available for editorial
use. A comprehensive press release with additional images and quotes
will follow.
Article Offline Attachments.
Daniel Brown wins the 36th BMW International OpenPDF, EN, 96.81 KB
Article Media Material.
My.PressClub Login
PressClub Streaming
RSS NEWS FEED.
With the PressClub RSS service, you can receive publications directly via news feed.Choose a topic-specific feed or use the general feed with the latest articles to stay up to date.
Click the link at the bottom of the website to find public RSS feeds, or use your personal RSS feed in the “Notifications” section of My.PressClub.
CO2 emission information.
Fuel consumption, CO2 emission figures and power consumption and range were measured using the methods required according to Regulation VO (EC) 2007/715 as amended. They refer to vehicles on the German automotive market. For ranges, the NEDC figures take into account differences in the selected wheel and tyre size, while the WLTP figures take into account the effects of any optional equipment.
All figures have already been calculated based on the new WLTP test cycle. NEDC figures listed have been adjusted to the NEDC measurement method where applicable. WLTP values are used as a basis for the assessment of taxes and other vehicle-related duties which are (also) based on CO2 emissions and, where applicable, for the purposes of vehicle-specific subsidies. Further information are available at www.bmw.de/wltp and at www.dat.de/co2/.
Nothing’s new Phone (3) is the company’s self-described first “true flagship,” a designation that, personally, I think is a mistake. But it also brings up the question, if the Nothing Phone (3) isn’t a “true flagship,” is Google’s Pixel?
This issue of 9to5Google Weekender is a part of 9to5Google’s rebooted newsletter that highlights the biggest Google stories with added commentary and other tidbits. Sign up here to get it delivered to your inbox early!
In my eyes, the Nothing Phone (3) is, in many ways, just another take on the formula Google has been using with the Pixel for the past few years. A high-end device that cares less about raw horsepower, and more about just delivering a focused and excellent experience. Google has had its missteps, certainly, but it’s working out pretty well.
Nothing Phone (3) feels as though it’s trying something similar.
Advertisement – scroll for more content
The Phone (3) isn’t as powerful as other high-end devices on paper, but it puts together a package that’s good at its job and more affordable than a lot of alternatives. Yet, I think Nothing has made a mistake here.
Nothing is marketing the Phone (3) as its first “true flagship.” That tagline has been used throughout teasers, the launch, and basically everywhere. Put simply, I don’t think that it’s quite true. Nothing is delivering, much like Pixel, a solid experience that puts the spec war aside, but it’s doing so in a way that takes away that “true” flagship title. Namely, the lower-resolution display and chipset are the main caveats here.
But, at the same time, Google also has a lesser chipset compared to the other flagships of the Android world. Pixel phones have been using Google’s custom Tensor chipset for the past few years, a chip that focuses on Google’s priorities over raw horsepower. Google’s approach, though, feels less compromising than Nothing’s to me, and it really just comes down to the marketing. Google is playing ball with high-end flagships, but it’s not demanding that title. Rather, Google quietly acknowledged for a long time that, no, it’s not trying to directly compete with the likes of “true” flagships by selling the Pixel series for a lower price. That cost has crept up in recent years as Google has closed that gap, though.
To me, I think of Pixels as flagship phones not because they have crazy specs, but because they’re the best Google has to offer, and are among the most capable smartphones you can buy. Flagships aren’t necessarily about what you’re getting on paper, but about the lack of compromise in the overall experience. If that weren’t the case, then you couldn’t call the base Galaxy S25 a flagship, for example.
Google’s formula for a flagship just happens to be a little different from Samsung, OnePlus, and some other Android brands.
Nothing, to its own right, can also do that. But demanding to be called a “true flagship” only invokes unnecessary expectations, and it has already done that before reviews even go live.
The “true flagship” schtick started before the chipset for Phone (3) was announced, leading almost everyone to think this device would be packing a Snapdragon 8 Elite. As soon as Nothing announced that the actual chip would be the considerably-less-powerful Snapdragon 8s Gen 4, a lot of that excitement immediately evaporated. Had Nothing never called this a “true flagship” and instead simply called it “our most powerful phone ever” or even recycled the tagline of Carl Pei’s previous venture, “flagship killer,” I think we’d see a lot more excitement out there. This feels like a “Never Settle” situation, where an overpromise is going to lead to Nothing’s own words coming back to bite them.
But what do you think? Is Nothing making a mistake with its branding? Are Google Pixel phones actually flagship devices? Let’s discuss!
This Week’s Top Stories
Nothing launch event
As mentioned, Nothing Phone (3) saw its official debut this week, alongside the new Nothing Headphone (1). See our coverage below:
Samsung leaks, get your Samsung leaks here!
As Samsung’s Unpacked event quickly approaches, the leaks keep on coming. There were a number of big, notable leaks this week, so see below for more:
Pixel 6a owners, you need to read this
Google has announced a new Pixel 6a update that will start rolling out on July 8. The update will include, for some users, a “battery management” feature that will have a major impact on battery life. See our coverage below for more information:
More Top Stories
From the rest of 9to5
9to5Mac: Apple is launching 15+ new products this fall, here’s what’s coming
9to5Toys: Nintendo Switch 2 Killswitch case is everything you can ask for, especially when the Joy-Cons stay attached
Electrek: Tesla launches Oasis Supercharger with solar farm and off-grid batteries
Follow Ben:Twitter/X, Threads, Bluesky, and Instagram
FTC: We use income earning auto affiliate links. More.
Malo Gusto has credited Chelsea’s mental strength for our run to the semi-finals of the FIFA Club World Cup but feels self-belief will also be vital as we enter the latter stages of the tournament.
The Blues continued our impressive from in the USA by defeating Palmeiras 2-1 in the quarter-finals, as Gusto’s cross was diverted in for an own goal late on, after Cole Palmer’s opening goal had been cancelled out by Estevao – who will join the Blues this summer.
It was far from plain sailing for the victors in Philadelphia. Following a strong 45 minutes by Chelsea, the Verdao fought back strongly in the second half, meaning the result was very much in the balance until that decisive moment seven minutes from full-time.
‘We are happy. It was our objective to get to this stage of the competition, it’s very important for us,’ said the Frenchman after the final whistle at Lincoln Financial Field.
‘It was a tough game again so we are very happy. They had a lot of energy during the game so we just had to deal with that, and that’s what we did.
‘I think during this tournament mental strength has been the most important thing and that’s why we are happy to be in the semi-final. We’ve shown that on the pitch in all the games we’ve had.’
Gusto was also pleased with his own crucial intervention, forcing the decisive error by Palmeiras when his cross took a touch of first defender Agustin Giay and then goalkeeper Weverton before finding its way into the back of the net.
‘It’s important to do something big, that’s what the coach asks of me on the pitch. So I just try to do my best to help the team as much as I can. That’s what I tried to do in this game and I’m happy we could win.
‘We play a lot of games so one game can be bad and then the second can be much better. The most important thing is to deal with that and to just stay focused on the team and trying to give our best, to make yourself better game by game. That’s what I try to do.’
Having already faced Flamengo in the group stage and Palmeiras in this quarter-final, Chelsea will now face Brazilian opposition for the third time at the 2025 Club World Cup, when we meet Fluminense in the semis. That game will kick-off at 8pm UK time on Tuesday 8 July.
Those previous experiences of facing teams from Brazil in the USA leave Gusto in no doubt that it will be another difficult challenge but, having come this far in the competition, his sights are firmly set on continuing our run all the way to the final.
‘The quarter-final was very tough, the Brazilian teams are doing very good during this competition,’ our full-back continued. ‘We will play against another one in the semi-finals, so we just need to be ourselves, dig deep and try to win.
‘It’s not enough yet, we still have a semi-final to play, to try to get to the final and try to win this trophy. Now we just have to believe in ourselves and to give our best.’
Tickets are available to see our Club World Cup semi-final against Fluminense in New Jersey and can be purchased from FIFA here. You can also watch the game live anywhere in world via DAZN by signing up for free here.
Grab your grill, your hot dogs, and your sparklers because it’s time to celebrate… Summer Games Done Quick 2025. The annual speedrunning charity marathon benefitting Doctors without Borders is back starting July 6th and finishes July 12th. Here’s how and most importantly what to watch as you enjoy the reason for the speedrunning season.
Summer Games Done Quick 2025 starts on Sunday July 6th at 1:30 PM ET on Twitch. Games Done Quick’s YouTube channel will have VODs up typically within a few short hours after the run completes so don’t worry if you miss anything live. There’s interesting runs every day during primetime so always check the schedule for what’s going on. You should also check in during off-hours and in the middle of the day because that’s how you find the hidden gems like LRock617’s run of Batman Forever: The Arcade Game at AGDQ 2025.
As for runs I’m excited about, there’s a Blue Prince race late Monday night July 7th, a Titanfall 2 race on Wednesday July 9th, and a Balatro run I will be seated for on Friday July 11th. For this year’s finale on Saturday, there will be a bonus run of Mario Kart World Knockout races and Deltarune Chapter 1 and 2. With the way I know GDQ works, I have a sneaking suspicion there will be a bonus incentive to donate towards that, if met, will unlock a run of Deltarune Chapter 3 and 4 as well. I hope I’m right.
With all the tumult going on in the video game industry right now it’s helpful to have something positive to turn to. That’s what Summer Games Done Quick is for. And this year, it’s needed more than ever as something to remind us that despite everything, video games, the people who make them, and the community, are pretty dang cool.
We may receive a commission on purchases made from links.
Gaming can be a pricey hobby, but in 2025, AMD is attempting to flip the script. The new Radeon RX 9060 XT delivers outstanding performance for just $350, offering 16GB of VRAM, strong 1080p and 1440p gaming performance, and even competitive ray tracing capabilities, all at a price mid-range buyers can actually afford for their rigs.
While recent years have seen GPU prices spike due to crypto booms, supply chain chaos, and AI demand, the RX 9060 XT marks a return to accessible, high-value graphics cards. It’s not just affordable, it’s efficient, future-ready, and stacked with real-world performance gains. Choosing the AMD Radeon 9060 XT 16GB was a no-brainer, as its closest competitor in the Nvidia GeForce RTX 5060 is a tough sell, with only 8GB of VRAM buffer — among other issues. Here’s why AMD’s GPU is the go-to choice for budget-conscious gamers in 2025 looking to make an upgrade.
Radeon RX 9060XT 16GB: AMD’s new budget champion
Unlike its bigger brother GPU, which is aimed at enthusiasts and 4K gaming, the RX 9060XT is a solid mid-ranger with excellent 1080p and strong 1440p performance. AMD offers the RX 9060XT in both 8 GB and 16 GB flavors, but given the high VRAM demands of the latest games like “Avatar Frontiers of Pandora,” “The Last of Us Part 1,” and various other titles, getting a 16GB version is the obvious choice. That is, unless you’re running a 1080p setup and don’t care for high-quality textures, ray tracing, or ultra settings.
The MSRP of GPUs is a sensitive topic, as in the recent past, they have been more like rough guidelines than actual maximum retail prices. That being said, at the time of writing, the RX 9060XT 16GB is available for purchase and can regularly be found around its MSRP of $350 for in-store pick-up at Micro Center, and $380-$390 on Amazon and Newegg.
With their 18-game average, Hardware Unboxed discovered that the 9060XT manages an impressive $5 per frame at 1440p at MSRP, the lowest of any card. Meanwhile, the Nvidia GeForce RTX 5060 8GB, RTX 5060 Ti 8GB, and RTX 5060 Ti 16GB average $5.45, $6.13, and $6.06 per frame, respectively. Intel’s ARC B580 is not included in this analysis, as it is unavailable at MSRP and faces compatibility issues in some games.
While AMD has traditionally trailed behind Nvidia significantly in ray tracing performance, the new RDNA 4 architecture marks a giant leap forward. The RX 9060XT comes with 32 second-gen ray accelerators and 64 AI accelerators. According to Tom’s Hardware, in a four-game average at 1440p with RT enabled, the 9060 XT manages an impressive average of 46.7 fps — just shy of the RTX 5060 Ti 16GB’s 51.5 fps. Even so, once MSRP is factored in, the AMD GPU comes out on top with a stronger $7.49 per ray-traced frame. Meanwhile, the 5060 Ti 16GB manages a $8.35 per frame, making it 11.5% more expensive than the Radeon graphics card.
More VRAM, Smarter Pricing, and Better Software
AMD’s Radeon RX 9060 XT stands out not just for performance, but for smart engineering choices. By sticking with GDDR6 memory, AMD delivers 16GB of VRAM — twice that of many similarly priced Nvidia cards — without inflating costs. That extra memory matters in today’s games, where high-res textures and larger frame buffers are the norm.
On the software front, AMD is catching up fast. FSR 4 introduces ML-powered upscaling similar to Nvidia DLSS, and the upcoming FSR Redstone (arriving late 2025) adds frame generation, neural radiance caching, and latency reduction, bringing AMD’s software ecosystem closer than ever to Nvidia.
Nvidia’s RTX 5060 Ti 8GB costs $380, offering half the VRAM for more money. The 16GB version is $430 — a $80 premium offering only marginal performance improvement over AMD’s 9060 XT, which is hard to justify at this price point. With more memory, lower power use, and fast-improving software, the RX 9060 XT isn’t just a good deal, it’s arguably the best value GPU you can buy today.
• Joined ESPN in 2009 • An FIA accredited F1 journalist since 2011
SILVERSTONE, England — Lando Norris emerged victorious in a wet and wild British Grand Prix that saw McLaren teammate Oscar Piastri stripped of his chance of victory by a 10-second penalty for a safety car infringement.
Piastri still claimed second place, but Norris’ home win, which marked his first at Silverstone and his fourth of the season, moved the British driver just eight points behind his teammate in the championship standings.
Nico Hulkenberg claimed the first podium of his 239-race career, bringing an end to the unwanted record of the driver with the most race starts without a podium.
Piastri, who passed Max Verstappen for the lead of the race on Lap 8 in wet conditions, was penalised 10 seconds for driving erratically ahead of a safety car restart on Lap 21.
The McLaren driver appeared to brake aggressively on the Hangar Straight, resulting in Verstappen, who was running in second place, passing him as the Red Bull was forced to take evasive action.
“Article 55.15 of the FIA Sporting Regulations required Car 81 [Piastri] to proceed at a pace which involved no erratic braking nor any other manoeuvre which is likely to endanger other drivers from the point at which the lights on the safety car are turned off,” a stewards statement said.
“What Car 81 did was clearly a breach of that article. In accordance with the penalty guidelines, we imposed a 10-second time penalty to Car 81.”
As the cars exited Stowe corner ahead of the restart, Verstappen spun to the inside of the track and dropped from second place to tenth when racing got back underway.
Verstappen’s spin moved Norris up to second place and meant the British driver only had to remain within 10-seconds of his teammate to ensure he took the lead when Piastri served his penalty at a pit stop on Lap 43.
The final round of pit stops saw the field move from the treaded intermediate tyres to slicks, with Piastri nearly losing control of his car as he took to the run-off area at Maggots on his outlap.
The Australian, who clearly felt the penalty was unfair, radioed his team to ask them to swap the cars back before the chequered flag, but with five laps left to go he was informed there would be no team orders to that effect.
Norris went on to take the victory by six seconds from Piastri, with Nico Hulkenberg’s Sauber 21 seconds behind in third and five seconds clear of Lewis Hamilton’s Ferrari in fourth.
An asteroid worth a whopping $10,000,000,000,000,000,000 that NASA is in the process of capturing could have shocking unintended consequences.
In 2023, the space agency announced that it was going to set off for the valuable asteroid, named 16 Psyche.
Thought to contain precious metals, including gold, iron and nickel, NASA is really keen to get its hands on the asteroid.
“Teams of engineers and technicians are working almost around the clock to ensure the orbiter is ready to journey 2.5 billion miles to a metal-rich asteroid that may tell us more about planetary cores and how planets form,” the space agency said in a statement released in July 2023.
The mission officially began in October 2023 as the spacecraft was launched from NASA’s Kennedy Space Center in Cape Canaveral, Florida.
Traveling at a speed of approximately 84,000mph through space, it’s expected to reach the valuable asteroid in August 2029.
While 16 Psyche may have been known about for a long time, experts are continuing to learn new things about the valuable asteroid all the time.
Newsweek reports that the projected value of the asteroid is 100,000 times the value of the world’s $100 trillion global domestic product due to the amount of gold, platinum and cobalt in it. That’s theoretically enough to make everyone on the planet a billionaire. Yikes.
Many have flocked to the comments section to express concerns that such an event could lead to gold becoming worthless, which in turn, would lead to a lot of people losing money.
One person said: “It wouldn’t make anyone billionaires but it will turn a lot billionaires to 0. Gold will lose its entire value.”
While a second added: “The price of gold would drop to a fraction of a penny an ounce, and nobody would become a billionaire from it. Simple supply and demand.”
A model of 16 Psyche at the Kennedy Space Center (CHANDAN KHANNA/AFP via Getty Images)
NASA estimates that this oddly shaped asteroid, which has a surface area of about 64,000 square miles (165,800 square kilometres), is made up of 30 to 60 percent metal.
It could also contain the exposed nickel-iron core of an early planet which is one of the building blocks of our solar system.
And if the asteroid’s materials really are worth $10 quintillion dollars, and that wealth was divided between every single living person, everyone would become very rich indeed.
There are some 8.062 billion humans alive, so dividing $10 quintillion dollars by our population would give us each a total of around $1.2 billion each.
I mean, it really is life changing stuff – so let’s see what happens in the coming years, eh?
