News

The Beats 360 headphones were just announced yesterday, and Amazon already has a whopping 80% (yes, 80%) discount on the official Beats 360 Cushion Kit.

• Preorder the Beats 360 headphones for $349

The Beats 360 are the first-ever customizable over-ear headphones, so the Cushion Kit lets you swap your earcups for a fresh set. Originally priced at $49.99, you can get the Beats 360 Cushion Kit for just $9.99 for a limited time. The Beats 360 and Cushion Kit are only available for pre-order, with a September 24 release date.

360 Cushion Kit for Earcups/Headband: was $49.99 now $9.99
This limited-time Amazon deal lets you score a whopping 80% discount on the Cushion Kit for the new Beats 360 headphones. Available in five fun colors, the cushion kit allows you to switch up your earcups and headband so you have a fresh set after a workout or run.

Preorder the Beats 360 headphones for $349View Deal

Our editor at large, Lance Ulanoff, got his hands on the Beats 360 and called them 'one of the most comfortable, best-sounding headphones I've ever worn'. The Beats 360 are the first-ever customizable over-ear headphones, letting you swap your ear cushions and headband cushion for fun colorways, including pastel pinks and blues, black, ivory, and vibrant yellow-green.

Designed for active lifestyles and everyday wear, the Beats 360 are the company's first-ever sweat- and water-resistant over-ear headphones. The cans deliver powerful, balanced sound with superb Active Noise Cancellation, a 1.75x improvement over the Beats Studio Pro.

In a Pier 57 auditorium filled with YouTubers of all ages and stripes, I felt a bit out of place. I haven't posted a full-length YouTube video to my channel in years, but listening to YouTube CEO Neal Mohan, other execs, and especially YouTube creators enthuse about new AI-boosted viewer and creator tools, I felt inspired again to, if not create, get back to my regular YouTube viewing habit.

The yearly Made on YouTube event, a confab on all the latest notable YouTube platform updates from creators and video consumers, featured a cavalcade of young, energetic YouTubers like Happy Kelli (Kelli Erdmann), Gremlita (Mina Le), and Derrick Gee, each of them introducing some fairly significant updates to the video platform, which Mohan claims has supported over half a million full-time jobs and contributed $60B to the US GDP in 2025.

Most notable for those who can't get enough of the creator videos and shorts are the new AI-powered Custom Feeds. While your current YouTube homepage is probably filled with subscriptions and algorithmic recommendations for your next watch, custom feeds are built via a prompt ("I want to see videos of tuxedoed cats playing jazz on pianos") and then live in a tab at the top of your feed. According to YouTuber Mina Lee, who gave the demo, the feeds constantly update with the latest videos that fit your request. Custom Feeds is expected on desktop and mobile this Fall (before the end of this year).

Just months after Google CEO Sundar Pichai unveiled "Ask YouTube" at Google I/O, YouTube is enhancing the AI-powered chatbot with stronger conversational skills and the ability to see inside videos. As we saw in the demo, you can go on quite a conversational journey with Ask YouTube, which can dig into the video's content and return time-stamped, contextual results, even pulling information from the comments.

Naturally, this journey could continue all the way to results with product specs and even the ability to buy a product directly. For now, the video creator who first talked about the product might miss out on affiliate revenue unless someone clicks from the Ask YouTube result to the original video. YouTube execs later told us they want to ensure that the affiliate connection still works.

Even if some of the AI tools YouTube is unveiling might appear to temporarily undermine creators, the new agentic editing tools look like a game changer.

Mobile App users will soon find "Ask Studio", a conversational editing assistant that lets you use natural language for edits. In the demo Happy Kelli ran, we saw how a series of prompts, each one building on the next without restating the initial task, turned a collection of clips into a complete video. She called it an "infinitely patient helper." YouTube says the tool will help creators edit a first draft, add typography, find the right music, and even sync it to the video action. And yes, it'll help you decide on the perfect hook: the thing people first see after the Thumbnail that might drag them into your video.

AI will also help build audio consumers a new "Your Podcast" lineup, which will basically be a short audio overview of podcasts you might like. YouTube creator Derrick Gee described it as "a personalized podcast about other podcasts, chosen for you — very meta."

YouTube Music is also getting its own chatbot, naturally called "Ask Music. Gee showed us how it can dig deep into the minutiae of an artist or song. In his example, he asked about the musical influences on Troye Sivan's new single She's the Best. One of the songs turned out to be Lou Reed's Walk on the Wild Side. As with other demos, the contextual journey can continue. In this case, Gee quizzed Ask Music about the bassist on Reed's song I(Herbie Flowers). Later, Sivan, who got his start on YouTube in 2007, joined Gee on stage (he did not sing).

Other updates include:

• YouTube Shorts being turned into Shorts Series with dedicated seasons, episodes, and automatic sequential playback
• GIFs in comments
• Custom thumbnails for different audiences
• Analytics adding rich insights and the ability to A/B test entire videos with live YouTube viewers
• Auto-dubbing for Live video, which is essentially realtime translation

Google Gemini's fingerprints are all over many of these updates, and I wondered how YouTube is navigating the choppy waters of AI sentiment.

Aparna Pappu, who leads YouTube Shorts, told me they're building these new tools and capabilities based on what creators need. Video creation remains, she said, a co-creative and collaborative process where the creator is always in charge. As for AI, "One of the ways [AI] serves us well: we only succeed when creators succeed.”

Amjad Hanif, VP of Creator Products, said that there was a time when creators were apprehensive about how AI might help them. "Now," he said, "it's hard to meet anyone who isn’t using it in their workflow."

• Microsoft has revealed a new Surface Pro 12-inch tablet and Surface Laptop 13-inch
• Both consumer devices are built around a 6-core Snapdragon X2 Plus processor
• They go on sale on October 13, 2026, starting at $1,149.99 / AU$1,999 for the Surface Pro and $1,199 / AU$2,099 for the Surface Laptop

It seems like Microsoft might be getting its mojo back for its Surface line-up of products, with Brett Ostrum, Corporate Vice President, Surface Devices, announcing at Qualcomm’s Snapdragon Summit that the company will be releasing a new Surface Pro 12-inch tablet and Surface Laptop 13-inch. Both will be powered by Qualcomm’s Snapdragon X2 Plus chips, which explains the venue choice. (I’m sure it had nothing to do with the fact that the Snapdragon Summit is held in Hawaii).

According to Microsoft, the two new Surface devices with Snapdragon X2 Plus CPUs will be up to 17% faster for Office workloads, up to 18% more battery efficient, and offer over 60% faster graphics performance.

As these are Microsoft products, there is also the now-customary boasting about AI performance, with the Qualcomm Hexagon NPU offering 80 TOPS (Trillion Operations Per Second) resulting in a claimed 95% faster local AI performance.

The Surface Pro 12-inch (2026) is a Windows 11 tablet built around the 6-core Snapdragon X2 Plus processor, partnered with either 16GB or 24GB of LPDDR5x RAM, and 256GB or 512GB storage. The 12-inch LCD display has a resolution of 2196 x 1464 and a refresh rate of up to 90Hz. The screen of this new model is, according to Microsoft, 25% brighter than the previous model, and there’s a new black color. Microsoft estimates around 15.5 hours of local video playback and 13 hours of web use on a single charge. It’ll start at $1,149.99 / AU$1,999 (around £900) and goes on sale on October 13, 2026.

The Surface Laptop 13-inch (2026) features the same 6-core Snapdragon X2 Plus chip and is available in the same memory and storage configurations as the Surface Pro. The 13-inch touchscreen has a resolution of 1920 x 1280 and a refresh rate of 60Hz, which I’m not going to lie, is a bit disappointing in a laptop these days, although Microsoft claims the 500-nit display is 25% brighter than the last model.

It also has two USB-C (USB 3.2) ports and a USB-A port, a 3.5mm headphone jack, Wi-Fi 7 and Bluetooth 5.4. Microsoft claims up to 22.5 hours of battery life, which if accurate is very impressive.

The Surface Laptop will start at $1,199 / AU$2,099 (around £920) and will also go on sale on October 13, 2026.

Microsoft’s announcement of two new consumer Surface devices is definitely welcome. Surface Pro tablets and Surface Laptops used to be Microsoft’s flagship products and showcases for cutting-edge Windows features. They certainly felt like Microsoft’s attempts to create a MacBook Air. The problem is, they never reached that potential, and when Apple switched to using its own Arm-based M series chips in 2020, the gap widened even further.

Surface devices also had a larger than life cheerleader in the form of Panos Panay, who had a major role at Microsoft and whose passion for the Surface line-up, that he helped create, was infectious (even if that infection didn’t translate into sales). After Panay left Microsoft for Amazon, it felt like the wind had been taken out of Surface’s sails, which wasn’t helped when Microsoft aimed recent Surface devices at commercial and business users, rather than mainstream consumers.

So, it’s good to see that these new Surface devices are firmly pitched at consumers, and the price points are around the same as the MacBook Air since Apple increased prices earlier this year. It also pits them against new Googlebooks, which are going on sale soon as well, and are priced similarly.

Will this mean Microsoft finally has a MacBook Air rival? While I’m glad it’s trying again, I’m not too sure. The Snapdragon X2 Plus is a decent mid-range chip, but the 6-core variant in the new Surface devices falls short of the 10-core CPU in the M5, especially as the M5 offers better single-core performance.

My main concern with the Surface Laptop is that the 1920 x 1280 screen just won’t compare to the display of the MacBook Air, which offers a 2560 x 1664 resolution for a much sharper image.

I’ve also seen the first five new Googlebook laptops, and they all come with higher resolution screens, and some also feature OLED.

The screen of the Surface Pro 12-inch (2026) is more promising, though if you want to use it as a laptop, you’ll need to buy a keyboard accessory, which adds to the price. However, it should still come in at under the price of Apple’s 13-inch iPad Pro.

So, I’m not entirely sure that these will be the MacBook Air/iPad Pro killers Microsoft needs them to be, but I’m glad Microsoft is at least trying again, and I could be proved wrong when we get the new Surface devices in for review soon.

• MORENA beat 26 tested models despite having only 1.5 billion parameters
• The model from Vambo AI was built specifically around African languages from scratch
• MORENA uses fewer tokens to represent African text than major rivals

Vambo AI has released MORENA, a 1.5B parameter language model covering 12 languages spoken across Africa, plus English, French and code.

The developers say it beats models from Google, Meta, Alibaba and HuggingFace on African language modelling and translation while being up to 8x smaller.

On a key benchmark, this LLM scored 1.408 bpb — the best of 26 models tested — beating its closest rival, which carried over five times as many parameters yet still scored only 1.423 bpb.

The model covers Nigerian Pidgin, Igbo, Yoruba, Hausa, Kiswahili, ChiShona, isiZulu, isiXhosa, Kinyarwanda, Setswana, Afrikaans, and isiNdebele.

Most projects serving these languages continue training an existing Llama variant, which forces them to keep a vocabulary meant for English and programming text.

Vambo AI instead fixed the tokenizer, the data mixture, and the language list before training began, after comparing several vocabulary sizes for cost and efficiency.

MORENA's vocabulary encodes African text with 1.39 times fewer tokens than Gemma 3 and 1.53 times fewer than Llama 3.2 on identical passages.

African text still costs 0.249 tokens for each byte against 0.234 for English, roughly 6% more, and the team cannot fully explain the difference.

The nearest competitor, Lugha-Llama-8B, an Africa-adapted Llama 3.1 variant, scores 1.423 bpb and loses to MORENA in eight languages out of 12.

The 12B Gemma system, by comparison, consumes roughly 11 times the compute of MORENA while producing a weaker score on the same text.

The chat-tuned instruct version scores 1.441 bpb, trailing Lugha-Llama-8B overall but leading it in five shared languages while using about 20% of its parameters.

The instruct model, after seeing three sample translations, renders English into five languages from Africa at 45.8 chrF++, statistically level with one dedicated translation system.

It trails a larger translation model by about 1.4 points, while general models of similar size typically land in a range from 9 to 14.

MORENA used 251.7B tokens during pretraining, followed by another 63B tokens during its mid-training stage.

Over 22,000 A100 GPU hours were reportedly required during development, with the associated computing resources valued at roughly $40,000.

According to the developers, this project was supported by UNDP, AIHub4SD, and CINECA, providing computing resources and other assistance during development.

This project also includes smaller releases, with versions containing 0.5B and 0.2B parameters available alongside the principal 1.5 billion model.

The 0.5B variants reportedly exceed every external system tested across 11 of the 12 languages covered by the project.

The 0.2B nano version is intended for speech-recognition rescoring, keyboard applications, and text normalisation tasks.

This nano version also costs 58 times less than Lugha-Llama-8B for each byte processed and generates 104 tokens each second on one A100 GPU.

A CPU-compatible build of the model also runs offline on a laptop, which suits settings where reliable GPU access is not guaranteed.

That gives the developers several model sizes for different computing requirements, rather than relying exclusively on the largest release.

MORENA also supports conversational applications, translation, and connections to other AI tools through its instruction-focused release.

Via Glitch Front

• Meta's glasses can no longer record if the capture LED is covered at any time, says independent test
• But the light is too hard to see if you're not very close, says German data protection agency
• EU bodies recommend actively getting full consent from anyone recorded, not assuming it

Detailed testing confirms that the Meta Glasses firmware update, designed to stop a potential privacy problem with surreptitious recording, does what it's supposed to do — but Germany's privacy enforcement agent says it's not enough, because the light itself is insufficiently visible.

The update was designed to address a simple exploit that bypassed the "no recording when the recording light is covered" feature.

In theory, that feature meant that it was impossible to secretly record someone or something by hiding the camera capture LED. The glasses could detect if the light is covered, and wouldn't record if that was the case.

In reality, it was immediately defeated by the kind of people who've helped earn the 'pervert glasses' label for the smart glasses: they discovered that the glasses only checked the light at the very beginning of recording, so you could simply start capturing, cover up the light and keep recording with the light now hidden.

The new update prevents them from doing that by re-checking the light every few seconds — but privacy watchdogs say it doesn't make much difference: even uncovered, the capture light on current models isn't good enough when it comes to privacy, consent or potential illegality.

In a detailed report by the Hamburg Commissioner for Data Protection and Freedom of Information, which involved a tear-down of the Ray-Ban Meta glasses and a detailed privacy analysis (via Notebookcheck.net), the Commissioner found that the LED light "is not easily visible in many environments, particularly outdoors and in direct sunlight". Meaning that even if you're using them as intended, that doesn't mean people can necessarily see (or recognize) the recording light

It's not just the lack of brightness. The beam angle is very narrow too, the report says, and that means you need to be pretty close and looking almost directly at the wearer to be able to see if the light is on.

The report also notes that the second-gen glasses behave differently to the first generation when the wearer asks about something or someone nearby. With the first generation, the LED "glows, very dimly". But in the second generation it doesn't glow at all.

The report argues that in most cases, if you're not recording friends or family only, then you can't assume informed consent. It recommends additional steps such as text notifications or signage, or actively speaking to the people you intend to film.

That ties in with a recent report by the French e-Enfance association, which reported this month that simply assuming someone saw the capture light does not necessarily mean they've consented to the storage, processing and distribution of their image.

A person needs to not only to have seen the light, but they must also understand what it indicates and be able to withhold their consent. And if minors are being filmed, e-Enfance says parents must provide permission.

It's wise to pay attention if you'll be filming in the country: French privacy laws are notably strict, with prison sentences and five-figure fines for deliberately invading someone's private life.

It's worth noting that these reports are based on European law, whereas US law is much more lenient: the US doesn't have the data protection and privacy protections members (and former members, such as the UK) of the European Union have, and its lawmakers are much more tech-friendly.

But no matter where you live, if you're a smart glasses wearer and could be recording others who aren't part of your close circle, it may be wise to ensure you can demonstrate that the people you're filming knew you were doing it, were able to opt out, and provided their consent.

There's a growing backlash against the use of smart glasses by various bad actors, and simply assuming consent, or even assuming that people are aware of what your glasses are doing just because they have a light, may not be enough to protect wearers from future legal consequences.

Try our TV size and model finder! You tell it how far you sit from your TV, we'll tell you what size to buy based on viewing angle advice from image quality experts, and we'll recommend our three top TVs at that size for different prices.

Dual-screen phones are having a moment (hello, iPhone Duo), but if you’re not yet ready to go full foldable, Xiaomi will meet you halfway: the Xiaomi 18 Pro comes with one-and-a-bit screens, and it’s shipping “globally” next year.

The successors to last year’s China-only Xiaomi 17 Pro, the 18 Pro and 18 Pro Max boast an extra display on their respective rear panels: a 2.66-inch OLED for the Pro and a 2.86-inch, 120Hz OLED for the Pro Max. These mini screens can display widgets, wallpapers, and media controls, and can be used to frame selfies taken with the phones’ rear cameras.

That’s similar to what was on offer with last year’s Xiaomi 17 Pro devices, though with this year’s phones, Xiaomi is introducing its own — and potentially even better — take on Samsung’s popular Privacy Display technology.

Like Privacy Display on the Galaxy S26 Ultra, Xiaomi’s Smart Privacy Display can restrict side-on viewing to prevent people from snooping on you at angles beyond 60 degrees. It can also work its magic on notifications and widgets without affecting the rest of the display.

Notably, though, Xiaomi’s take on the technology can be applied to apps used in split-screen mode (maybe you only want to obscure one app, and not the other), as well as to specific locations and app notifications (because some pop-ups may be more risqué than others).

Smart Privacy Display is a feature of both the Xiaomi 18 Pro and 18 Pro Max, though the latter obviously has a larger screen: it’s 6.9 inches vs. 6.4 inches on the Pro.

I got the chance to get hands-on with the smaller Pro model at Snapdragon Summit 2026 (thanks, Qualcomm, for flying me out to Hawaii), and although I wasn’t able to unlock the display unit to test Smart Privacy Display, the rear OLED display was partially functional.

I used it to snap a few selfies using the Xiaomi 18 Pro’s 200MP main lens, which can be activated with a wave gesture or a press of the volume button.

As for the size and weight of the device, it’s (very) similar to the iPhone 18 Pro, which I suspect is a deliberate choice on Xiaomi’s part. Lest we forget that the iPhone Duo-like Xiaomi 18 Fold launched exactly one day before the iPhone Duo last month.

Other specs include the new Snapdragon 8 Elite Gen 6 and Snapdragon 8 Elite Extreme Gen 6 chipsets for the Pro and Pro Max, respectively, as well as a shared triple-lens camera array comprising a 200MP main lens, a 200MP telephoto lens, and a 50MP ultra-wide lens.

The Pro comes equipped with a 7,000mAh silicon-carbon battery, while the Pro Max gets a larger 8,500mAh cell. Those capacities may be smaller outside of China, though.

We also don’t have much clarity on what Xiaomi means by a “global” launch. As mentioned, neither the Xiaomi 17 Pro nor 17 Pro Max launched outside of China, but the Xiaomi 17 Ultra did arrive in the UK and Europe, so I'd wager that those regions will get at least one of the Xiaomi 18 Pro and 18 Pro Max at some point. A US or Australia launch for either phone seems unlikely, though.

As for price, the 18 Pro will start at ¥5,999 CNY (around $900 / £675 / AU$1,300), and the 18 Pro Max at ¥6,999 (around $1,050 / £785 / AU$1,500). If and when these phones do arrive outside of China, though, their respective prices are likely to be much higher.

A new NYT Connections puzzle appears at midnight each day for your time zone – which means that some people are always playing 'today's game' while others are playing 'yesterday's'. If you're looking for Wednesday's puzzle instead then click here: NYT Connections hints and answers for Wednesday, September 23 (game #1200).

Good morning! Let's play Connections, the NYT's clever word game that challenges you to group answers in various categories. It can be tough, so read on if you need Connections hints.

What should you do once you've finished? Why, play some more word games of course. I've also got daily Strands hints and answers and Quordle hints and answers articles if you need help for those too, while Marc's Wordle today page covers the original viral word game.

SPOILER WARNING: Information about NYT Connections today is below, so don't read on if you don't want to know the answers.

Today's NYT Connections words are…

• THE BEAR
• FOAM FINGER
• POGO STICK
• FACE PAINT
• BOBBLEHEAD
• BEER HAT
• UNCLE SAM
• CAPE
• CREATION OF
• ADAM
• TRAMPOLINE
• FANGS
• BREATHE
• FAKE BLOOD
• I'M WITH STUPID
• SHIRT
• SNAKES IN A CAN
• HERB TEA

What are some clues for today's NYT Connections groups?

• YELLOW: Dracula disguise
• GREEN: Bouncy items
• BLUE: It could be you
• PURPLE: Letters in common

Need more clues?

We're firmly in spoiler territory now, but read on if you want to know what the four theme answers are for today's NYT Connections puzzles…

What are the answers for today's NYT Connections groups?

• YELLOW: PARTS OF A VAMPIRE COSTUME
• GREEN: THINGS WITH SPRING/S
• BLUE: FEATURING POINTING
• PURPLE: ANAGRAMS

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

The answers to today's Connections, game #1201, are…

• YELLOW: PARTS OF A VAMPIRE COSTUME: CAPE, FACE PAINT, FAKE BLOOD, FANGS
• GREEN: THINGS WITH SPRING/S: BOBBLEHEAD, POGO STICK, SNAKES IN A CAN, TRAMPOLINE
• BLUE: FEATURING POINTING: CREATION OF ADAM, FOAM FINGER, I'M WITH STUPID SHIRT, UNCLE SAM
• PURPLE: ANAGRAMS: BEER HAT, BREATHE, HERB TEA, THE BEAR

• My rating: Hard
• My score: 1 mistake

I got away with only making one mistake today, but it could have easily been more — as I yet again felt utterly baffled by the 16 tiles.

My mistake was linking FOAM FINGER, FACE PAINT, BEER HAT and BOBBLEHEAD, thinking they all had something to do with being a sports fan — not that I actually know what a BOBBLEHEAD is, except that it sounds like tat.

After this, PARTS OF A VAMPIRE COSTUME was the only group I felt confident in. For THINGS WITH SPRING/S I felt TRAMPOLINE and POGO STICK had bouncing in common and took a guess with the other two, while FEATURING POINTING I thought was all about comedy T-shirts.

• YELLOW: THINGS YOU GRAB BEFORE LEAVING THE HOUSE: KEYS, PHONE, SUNGLASSES, WALLET
• GREEN: WEBSITE HURDLES: AGE VERIFICATION, CAPTCHA, COOKIE CONSENT, PAYWALL
• BLUE: THINGS THAT COME IN CANS: CONDENSED MILK, PAINT, SHAVING CREAM, SPAM
• PURPLE: ENDING IN SYNONYMS FOR "TERM": BITCOIN, RECORD LABEL, RUB-A-DUB, SURINAME

NYT Connections is one of several increasingly popular word games made by the New York Times. It challenges you to find groups of four items that share something in common, and each group has a different difficulty level: green is easy, yellow a little harder, blue often quite tough and purple usually very difficult.

On the plus side, you don't technically need to solve the final one, as you'll be able to answer that one by a process of elimination. What's more, you can make up to four mistakes, which gives you a little bit of breathing room.

It's a little more involved than something like Wordle, however, and there are plenty of opportunities for the game to trip you up with tricks. For instance, watch out for homophones and other word games that could disguise the answers.

It's playable for free via the NYT Games site on desktop or mobile.

Meta Connect 2026 is busy. Alongside the next-generation Meta Glasses, a camera-free version of the Ray-Ban Meta Glasses, and a new collaboration with Korean pop star Lisa, Meta is adding a significant new capability to its glasses: hearing enhancement.

The US Food and Drug Administration (FDA) has cleared the feature, which turns a pair of Meta Glasses into a hearing aid for people with mild to moderate hearing loss. It amplifies sound based on your hearing profile and uses machine learning to boost voices and suppress background noise.

Meta isn't the first here, though. Apple's AirPods Pro 2 are already cleared as over-the-counter hearing aids and Nuance offers a pair of smart shades with a similar function. Meta, though, is betting on its form factor: glasses sit outside the ear, and many people already wear them all day.

It's not free, though. The feature launches with a free 30-day trial, after which it costs a one-time payment of $149.99, or comes as part of a Meta One subscription. In the US, it's also FSA/HSA eligible, with Meta saying it will share more details closer to the launch, which in the US is set for later in 2026. Even including the price of compatible glasses, Meta says it's significantly cheaper than traditional hearing aids.

The feature is launching in the US first, with Meta planning to expand to other countries as it works through each region's medical device regulations.

I spoke with Ravish Mehra, Director of Research Science at Meta, about why Meta thinks glasses are the right home for hearing enhancement. Mehra, who has spent nearly 12 years at the company, says the work started more than five years ago.

"We have a very unique audio platform that has microphone arrays, that has glasses, is socially acceptable in terms of people wearing glasses, and we have onboard compute," he says.

According to Mehra, one in six Americans have mild to moderate hearing loss, yet only around 15% of them do anything about it. He notes that hearing aids are expensive, uncomfortable to wear all day, and signal to others that you have a hearing impairment. So people "put the issue under the rug," Mehra says, telling themselves, "Okay, it's just people are mumbling, and that's why I'm not able to hear."

Glasses sidestep the stigma in his view: "Nobody else can tell just by looking at your glasses that you have hearing impairment.”

"Every person's hearing loss is slightly different," Mehra says. Instead of a clinic visit and multiple fittings, Meta's hearing test takes around one to two minutes with your phone, through the Meta AI app, and your glasses. You listen to sounds at different frequencies, indicate when you can barely hear them, and the system builds a hearing profile showing where you need amplification.

"You don't have to go anywhere to get these glasses fit for your hearing profile," Mehra says. "You just do a very simple test at home, and whenever your hearing needs change, you just redo the test."

That test was central to the FDA clearance. Meta tested the feature with a group of people with mild to moderate hearing loss, and its results showed the phone-based test was equivalent in performance to a professional fitting, within an acceptable margin.

"We are actually the first self-fitting hearing aid on glasses in the market," Mehra says.

The glasses' six microphones capture sound around you — currently being used on Meta Glasses — which is then amplified where your profile shows a deficit. That also amplifies noise, so a machine learning model trained to recognize voices brings the background down. It all runs on-device, and speed is critical. "All of this has to happen in a very short latency, because otherwise you start hearing echoes or start hearing distortions," Mehra says.

There are two modes: 360 mode amplifies sound all around you, while Focus mode suits environments like a restaurant, where you want to hear the person across the table.

"We actually use the microphones to first do beamforming, which is like focusing in the forward direction more," Mehra explains. "And then […] we train the system to enhance the voices while not enhancing or suppressing the noise."

The frame helps, too. "Because the glasses have a much wider baseline, you can put the microphones much farther away," says Mehra. This makes it easier to isolate speech from noise than with microphones placed millimeters apart in the ear.

I got a brief demo: Meta had us put in earplugs to simulate hearing loss, then wear the glasses. They did properly amplify the people directly around me, and even those a bit farther away. Focus mode noticeably singles out the person in front of you, but it wasn't perfect, and of course earplugs are no substitute for real hearing loss.

Amplifying sound isn't the hardest part, though. "One of the issues that any kind of real-time amplification feature faces is called feedback," Mehra says. "When you amplify something through the speakers, they feed back into the microphones."

On glasses, with their open-ear design, nothing separates the speakers from the microphones. Meta's system has to detect feedback before it builds, suppress it, and still deliver enough amplification. "If you're not able to cancel that feedback, then you hear these very sharp noises," Mehra says.

Then there's processing, which is added on top of a broader feature set the glasses offer, so Meta had to squeeze the feature into what Mehra calls "a very small envelope" of onboard compute power.

Meta hasn't confirmed which glasses will support it, only that they must meet a minimum performance bar, or confirmed a rollout schedule other than late 2026. Given the processing involved, I'd be surprised to see it on older models like the first-gen Ray-Ban Meta glasses; all of the demos at Meta Connect have been on the more recently released Meta Glasses.

Meta isn't just trying to make its glasses hear for you; it's trying to help you hear, too. It's a new category for the company, but Mehra is in no doubt about the bet, telling me: "We strongly feel like glasses is the ideal form factor."

Even as public opinion turns against camera-equipped smart glasses, Meta is pushing forward with its generally well-received and quite popular Ray-Ban Meta line. Alongside the much-anticipated Meta Ray-Ban Audio glasses, which skip cameras entirely and come in two styles, Clubmaster and Burbank, there's also the next generation of the Ray-Ban Meta glasses.

Like the previous generation, the Ray-Ban Meta (Gen 3) still features onboard cameras with the all-important glowing LED ring that alerts you when a photo is being taken or a video is being recorded. It's an important safeguard, and Meta has added more protections against folks who try to disable or conceal it, but concerns remain.

Meta's pushing forward with the Gen 3, even if the audio-only models prove more popular. They're launching today in three styles: Ray-Ban Meta Aviator, Ray-Ban Meta Zena, and a new version of the Ray-Ban Meta Wayfarer. Pricing starts at $449 in the United States; we're working to confirm UK and Australia pricing.

The story across the board is better hardware, all tucked into smart glasses that more closely resemble real glasses, with lighter and thinner builds where possible. The Aviators are particularly exciting to see adopted for the AI and smart era. The frames are thicker than the classic wire ones, but they still frame the face well. Zena is decidedly modern and sleek as a cat's eye, while Wayfarer is still the iconic, cool frame that Ray-Ban introduced years ago.

I'm testing the Ray-Ban Meta Aviator in a brown shade, and you can see my unboxing video embedded below. I like this color, especially compared with my speckled brown Gen 1 Ray-Ban Meta Wayfarers.

There are 27 total combinations when you account for colors and lens types, and if you're willing to pay for it, you can fit these with transition lenses and even prescription lenses.

Regardless of frame style or lens type, you benefit from improved tech onboard. For starters, the cameras — if you choose to use them, and as mentioned, the ring will glow when you do — can take 12-megapixel photos and record 3K Ultra HD video, with Dolby Atmos support for a more cinematic audio experience during playback. There's also a Dynamic Photo mode, similar to Live Photos on iOS, which snaps a few shots so you can pick the best one.

Battery life is also improved. It isn't quite at the 12 hours of the new Meta Ray-Ban Audio models, but the Gen 3s last up to nine hours on a full charge. That's an hour longer than the Gen 2s, and the included charging case lets you recharge on the go.

Finally, just like the Meta Glasses — including the Kylie edition — that launched earlier in 2026, the Gen 3s get more microphones. Six of them improve voice pickup, whether you're drafting a message, asking Meta AI a question, or on a call. Meta promises they cut over 90% of background noise, and we look forward to putting that to the test.

While it's not a night-and-day difference over the second generation, the Ray-Ban Meta (Gen 3) shows that Meta has settled on the core feature set it wants for its AI glasses. Now, in a move likely to broaden their appeal, it's focusing on offering more styles, and its partnership with Ray-Ban is clearly still evolving. The Gen 3s continue the quest for smart glasses that feel closer to the regular versions of these frames, while adding the latest tech for a compelling experience.

Just how compelling will be weighed against the added privacy concerns, but let's hope Meta keeps adding more privacy features and keeps cracking down on those who try to block out the LEDs.

We'll be back soon with our full thoughts — and if you ask us, the Aviators look pretty, pretty nice — but if you're already sold, the Ray-Ban Meta (Gen 3) are up for order right now, starting at $449 in the US, from Meta, Ray-Ban, and other authorized sellers.