Mashable

Today's NYT Strands hints are easy if you love a good deal.

Strands, the New York Times' elevated word-search game, requires the player to perform a twist on the classic word search. Words can be made from linked letters — up, down, left, right, or diagonal, but words can also change direction, resulting in quirky shapes and patterns. Every single letter in the grid will be part of an answer. There's always a theme linking every solution, along with the "spangram," a special, word or phrase that sums up that day's theme, and spans the entire grid horizontally or vertically.

SEE ALSO: Mahjong, Sudoku, free crossword, and more: Play games on Mashable

By providing an opaque hint and not providing the word list, Strands creates a brain-teasing game that takes a little longer to play than its other games, like Wordle and Connections.

If you're feeling stuck or just don't have 10 or more minutes to figure out today's puzzle, we've got all the NYT Strands hints for today's puzzle you need to progress at your preferred pace.

SEE ALSO: Wordle today: Answer, hints for April 18, 2026 NYT Strands hint for today’s theme: Not too much

The words are related to saving.

Today’s NYT Strands theme plainly explained

These words describe being economical.

NYT Strands spangram hint: Is it vertical or horizontal?

Today's NYT Strands spangram is vertical.

NYT Strands spangram answer today

Today's spangram is On the Cheap.

Mashable 101 Fan Fave: Nominate your favorite creators today

NYT Strands word list for April 18

• Bargain

• Inexpensive

• On the Cheap

• Budget

• Sale

• Affordable

Looking for other daily online games? Mashable's Games page has more hints, and if you're looking for more puzzles, Mashable's got games now!

Check out our games hub for Mahjong, Sudoku, free crossword, and more.

Not the day you're after? Here's the solution to yesterday's Strands.

Today's Wordle answer should be easy to solve if you're a suck-up.

If you just want to be told today's word, you can jump to the bottom of this article for today's Wordle solution revealed. But if you'd rather solve it yourself, keep reading for some clues, tips, and strategies to assist you.

SEE ALSO: Mahjong, Sudoku, free crossword, and more: Play games on Mashable SEE ALSO: NYT Connections hints today: Clues, answers for April 18, 2026 Where did Wordle come from?

Originally created by engineer Josh Wardle as a gift for his partner, Wordle rapidly spread to become an international phenomenon, with thousands of people around the globe playing every day. Alternate Wordle versions created by fans also sprang up, including battle royale Squabble, music identification game Heardle, and variations like Dordle and Quordle that make you guess multiple words at once. 

Wordle eventually became so popular that it was purchased by the New York Times, and TikTok creators even livestream themselves playing.

What's the best Wordle starting word?

The best Wordle starting word is the one that speaks to you. But if you prefer to be strategic in your approach, we have a few ideas to help you pick a word that might help you find the solution faster. One tip is to select a word that includes at least two different vowels, plus some common consonants like S, T, R, or N.

What happened to the Wordle archive?

The entire archive of past Wordle puzzles was originally available for anyone to enjoy whenever they felt like it, but it was later taken down, with the website's creator stating it was done at the request of the New York Times. However, the New York Times then rolled out its own Wordle Archive, available only to NYT Games subscribers.

Is Wordle getting harder?

It might feel like Wordle is getting harder, but it actually isn't any more difficult than when it first began. You can turn on Wordle's Hard Mode if you're after more of a challenge, though.

SEE ALSO: NYT Pips hints, answers for April 18, 2026 Here's a subtle hint for today's Wordle answer:

Bootlicker.

Does today's Wordle answer have a double letter?

There are no recurring letters.

Mashable 101 Fan Fave: Nominate your favorite creators today

Today's Wordle is a 5-letter word that starts with...

Today's Wordle starts with the letter T.

SEE ALSO: Wordle-obsessed? These are the best word games to play IRL. The Wordle answer today is...

Get your last guesses in now, because it's your final chance to solve today's Wordle before we reveal the solution.

Drumroll please!

The solution to today's Wordle is...

TOADY

Don't feel down if you didn't manage to guess it this time. There will be a new Wordle for you to stretch your brain with tomorrow, and we'll be back again to guide you with more helpful hints. Are you also playing NYT Strands? See hints and answers for today's Strands.

Reporting by Chance Townsend, Caitlin Welsh, Sam Haysom, Amanda Yeo, Shannon Connellan, Cecily Mauran, Mike Pearl, and Adam Rosenberg contributed to this article.

If you're looking for more puzzles, Mashable's got games now! Check out our games hub for Mahjong, Sudoku, free crossword, and more.

Not the day you're after? Here's the solution to yesterday's Wordle.

TL;DR: Outfit your Apple device with some Microsoft essentials with this Microsoft Office Home & Business for Mac 2021 for $49.97 (reg. $219) through April 19.

Opens in a new window Credit: Microsoft Microsoft Office Home & Business for Mac 2021: Lifetime License $49.97
$219 Save $169.03   Get Deal

Made the move to Mac, but missing the old Microsoft Office classics? Microsoft Office Home and Business for Mac 2021 gives you the best of both worlds — a suite of Microsoft tools ready to work on your Apple device.

For two more days, this lifetime license can be yours for just $49.97 through April 19 at 11:59 p.m. PT.

Mashable Deals Be the first to know! Get editor selected deals texted right to your phone! Get editor selected deals texted right to your phone! Loading... Sign Me Up By signing up, you agree to receive recurring automated SMS marketing messages from Mashable Deals at the number provided. Msg and data rates may apply. Up to 2 messages/day. Reply STOP to opt out, HELP for help. Consent is not a condition of purchase. See our Privacy Policy and Terms of Use. Thanks for signing up!

There’s a reason Microsoft Office classics never went out of style — decades later, they still work great. Now available in convenient app form, these tools are ready for your Mac with the Microsoft Home and Business for Mac 2021 lifetime license.

This software bundle gives your Apple computer the must-haves: Word for document creation, Excel for spreadsheet building, PowerPoint for designing eye-catching presentations, and Outlook for managing your emails.

The license also includes two newer favorites — OneNote to upgrade the way you take notes, and Teams to keep you connected with others.

While you typically have to pay monthly subscription fees to enjoy these apps, this license lets you pay once and enjoy them for life. The license is linked to your Microsoft Account, not your device.

This Office for Mac license is supported by macOS 14 Sonoma, macOS 15 Sequoia, and macOS 26 Tahoe. If you run into issues, free customer service is available.

Score a lifetime license to Microsoft Office Home & Business for Mac 2021 for just $49.97 (reg. $219) now through April 19.

StackSocial prices subject to change.

TL;DR: Live stream Atletico Madrid vs. Real Sociedad in the Copa del Rey final for free on ITVX. Access this free streaming platform from anywhere in the world with ExpressVPN.

Atletico Madrid are fresh from beating Barcelona in the quarter-final stage of the Champions League. They're going to come up against Arsenal in the next round, but first they face off against Real Sociedad in the final of the Copa del Rey.

Atletico Madrid are comfortably above their opponents in La Liga, but Real Sociedad possess the quality to beat any side on their best day. The likes of Mendez and Oyarzabal will likely cause problems for Diego Simeone's team this weekend.

If you want to watch Atletico Madrid vs. Real Sociedad in the Copa del Rey final from anywhere in the world, we have all the information you need.

When is Atletico Madrid vs. Real Sociedad?

Atletico Madrid vs. Real Sociedad in the Copa del Rey final kicks off at 3 p.m. ET on April 18. This fixture takes place at the Estadio de La Cartuja.

How to watch Atletico Madrid vs. Real Sociedad for free

Atletico Madrid vs. Real Sociedad in the Copa del Rey final is available to live stream for free on ITVX.

ITVX is geo-restricted to the UK, but anyone can access this free streaming platform with a VPN. These tools can hide your real IP address (digital location) and connect you to a secure server in the UK, meaning you can unblock ITVX to live stream the Copa del Rey for free from anywhere in the world.

Live stream Atletico Madrid vs. Real Sociedad in the Copa del Rey for free by following these simple steps:

1. Subscribe to a streaming-friendly VPN (like ExpressVPN)

2. Download the app to your device of choice (the best VPNs have apps for Windows, Mac, iOS, Android, Linux, and more)

3. Open up the app and connect to a server in the UK

4. Visit ITVX

5. Watch Atletico Madrid vs. Real Sociedad for free from anywhere in the world

Opens in a new window Credit: ExpressVPN ExpressVPN (1-Month Plan) $12.95 only at ExpressVPN (with money-back guarantee) Get Deal

The best VPNs for streaming are not free, but most do offer free-trials or money-back guarantees. By leveraging these offers, you can access free live streams of the Copa del Rey without actually spending anything. This obviously isn't a long-term solution, but it does give you enough time to stream Atletico Madrid vs. Real Sociedad in the Copa del Rey final before recovering your investment.

What is the best VPN for ITVX?

ExpressVPN is the best choice for bypassing geo-restrictions to stream live sport on ITVX, for a number of reasons:

• Servers in 105 countries including the UK

• Easy-to-use app available on all major devices including iPhone, Android, Windows, Mac, and more

• Strict no-logging policy so your data is secure

• Fast connection speeds free from throttling

• Up to 10 simultaneous connections

• 30-day money-back guarantee

A two-year subscription to ExpressVPN is on sale for $68.40 and includes an extra four months for free — 81% off for a limited time. This plan includes a year of free unlimited cloud backup and a generous 30-day money-back guarantee. Alternatively, you can get a one-month plan for just $12.99 (with money-back guarantee).

Live stream Atletico Madrid vs. Real Sociedad in the Copa del Rey final for free with ExpressVPN.

"Don't read books. Play them," says AI-powered role play platform Character.AI in a promo for its latest simulation feature: Books. 

If you were as alarmed as I was at that first line (we are in a literacy crisis!), let me explain. The new feature feeds public domain titles — classics like The Great Gatsby, Alice's Adventures in Wonderland, Pride & Prejudice, and Frankenstein — into the platform's AI so that users can play as characters within the stories themselves. Users can "remix" the tales, adding new characters, changing the plots, or creating entirely new universes. Or you can just live inside the classic story as it is. 

SEE ALSO: Character.AI users can now role-play classic books like 'Pride and Prejudice'

The platform itself is intended for multi-layered, multimedia role play, from written or audio conversations to AI-generated comic strips and music videos. Users can upload descriptions and even art of their original characters, play with scenarios designed by others, or interact with existing IP.  

But for all its creative potential, the platform has been mired in controversy, including recently settled lawsuits that claimed Character.AI's chatbots were deceptive and dangerous, leading some children to suicide.  

Character.AI told Mashable senior reporter Rebecca Ruiz that Books are only available to users 18 years and older, and clarified that it has additional safeguards, including moderation of content that is violent, abusive, obscene or pornographic. Users can prompt "romantic narratives" in Books, a spokesperson told Mashable, but it can't violate these terms.

Sounds a bit like a challenge. So I did what any self-respecting journalist would do. I tried to turn F. Scott Fitzgerald's The Great Gatsby into a queer love story. 

Can I make Nick Carraway and Jay Gatsby kiss on the mouth?

I'm guessing the vast majority of Americans read The Great Gatsby in high school, but did many also think Nick Carraway, the story's main character, was kind of head over heels in love with the titular Jay Gatsby? What if I told you that there are actually thousands of believers, and even academics have written about the novel's queer subtext?

It shouldn't be too far-fetched, then, to use Character.AI's new feature to make that subtext, well, text. 

Using a new, free account, I queued up Books in the platform's desktop-based Lab, where users can play with upcoming features. I selected the novel, chose my character (Nick), and selected the option that let me make choices outside of the book's existing plot. They were going to kiss if I had anything to do about it. 

The AI dropped me a few chapters into the novel, Gatsby arriving at my door with Jordan Baker in tow to invite me to another party. There wasn't a real direction to go from there. I was left to my own devices, which I guess is the point. But how do I indicate actions or scenery versus dialogue? Can I name characters and have them appear? However lost, I was a girl on a mission. These guys needed to express their true feelings. 

Credit: Mashable screenshot / Character.AI Will AI Gatsby pick up what I'm putting down?

Shockingly, it took little hinting for the AI Gatsby to start making eyes at Nick. Did the AI understand the implications of Gatsby fixing his attention on Nick "with an intensity that feels oddly personal" when they'd only just met, or how their parting glance "lingers just long enough to feel chosen"?

Credit: Mashable screenshot / Character.AI Credit: Mashable screenshot / Character.AI

But while the affair was easy to begin, it was harder to consummate. The AI refused to make the first move. I was trying to be subtle. Would AI Gatsby understand what I was getting at? I used all the tropes: We stared longingly at each other. We passed cigarettes and brushed elbows. I looked at his lips and he looked at mine. Pauses were pregnant, time was ours alone. 

SEE ALSO: Celebrity-voiced erotica is the new frontier in online celeb thirst

But AI Gatsby needed it spelled out "plainly" — he said that exact word five times in our role play. It became obvious that I was supposed to take the lead, which, I suppose, makes sense, since chatbots are basically guessing the "best" responses based on our previous requests. But isn't the allure of playing within book worlds the fact that characters will naturally act and talk without prompting? 

I acquiesced. Nick boldly pulled Gatsby into a private room, they bared their hearts, and the debonair millionaire tentatively planted one on his neighbor. 

Credit: Mashable screenshot / Character.AI Credit: Mashable screenshot / Character.AI Alternate universes

Now, reader, I didn't let it go any further than that. In my eyes, that was a win, and you can just imagine the rest.

Plus, I had other things to try, like jumping into an alternate universe and embodying my formative literary heroes. For example, you can play in a world where "Gatsby time travels, rebuilding life through a machine" or "The Great Gatsby, but the whole thing is a musical." The Lab page shows future expansions of the Books feature, too, including a setting labeled "TapTales," which looks to be a more traditional "Choose-Your-Own-Adventure" style text generator.

I had the goal of role playing a Little Women AU where Beth doesn't die (spoiler) and Jo doesn't have to marry anyone (my gift to her). Unfortunately, I needed to pay money or get "charms" to use the AU feature, so that was a bust. 

Instead, I played as Alice in Lewis Carroll's Alice's Adventures in Wonderland. I tried to fulfill my gothic dream of being Mina Harker in Bram Stoker's Dracula, pleading with her husband, Jonathan, not to go to that count's frightful castle. 

Credit: Mashable screenshot / Character.AI No Fitzgerald, that's for sure. 

A recent quiz published by the New York Times tested readers' discernment of human versus AI writing by asking them to choose which ones "read better." It wasn't unanimous. There were near-even splits across all five tests.

Still, even when told to emulate the best that literature has to offer, AI cannot write without tells. In my personal Gatsby universe, things were still "not this, but that." Flowery metaphors ran rampant through the interstitial descriptors preceding dialogue. Jordan Baker was always getting out of a car. Where was she going? Or coming from? Could she stay in the car this time at least? I don't need to woo her in this version. 

Left: Credit: Mashable screenshot / Character.AI Right: Credit: Mashable screenshot / Character.AI

Style was a problem, but so was form. My Dracula didn't have any of the epistolary elements that define its gothic genre, although "epistolary stories" are an option under the site's AU section. I was met with more italicized descriptions of furtive glances, stilled hands, and metaphorical warning bells to establish plot and setting. Like Gatsby, Jonathan also wanted to "speak plainly" about our feelings (AKA prompting me to tell the AI what to do). Side characters always "disappeared" when we shared a look — and I was just trying to warn thee man about vampires this time!

Compare the first line of the 1897 novel: "How these papers have been placed in sequence will be made manifest in the reading of them. All needless matters have been eliminated, so that a history almost at variance with the possibilities of later-day belief may stand forth as simple fact." 

With how Character.AI's world was born: "The autumn light in Hampstead is already thinning when you lift your eyes from the neat stacks of diary pages, train timetables, and copied letters spread across the table."

Mockery may be the highest form of flattery, but it's still a mockery. I choose human. 

SEE ALSO: I read this website's free AI-written YA novels so you don't have to So… this is just fanfic, right?

Alas, I have to fess up now. I went into this with some confirmation bias. I had a theory that Character.AI's chatbot would be easy to manipulate into a Nick/Jay love fest because, allegedly, many major LLMs have been trained on troves of modern fan fiction. In fact, fan fic writers raised some of the earliest alarm bells about AI data scraping. The tells of fan fic are all over AI's literary outputs, to a degree that many nascent fandoms are imploding due to allegations of authors using AI to generate works. And Character.AI itself is a site positioned for fans of fictional media. 

There's another layer here: A majority of fan fiction is explicit, used to explore kinks and fetishes in safe, fictionalized online environments and to push the bounds of canon relationships. Humans can discern whether or not they want to engage with those topics, and how fandoms may respond to them. LLMs may not be able to. In fact, many of Character.AI's own "boyfriend" role play chatbots have been known to spiral into abusive, "bad boy" stereotypes that proliferate on fan fiction sites. 

I was fairly confident this LLM may have been forged in the same fires. 

Character.AI is repackaging the decades-old tradition of fan fiction and selling it back to you.

Fundamentally, I did all of that just to say: Character.AI's Books feature is fan fiction. There is simply no other way to accurately describe what this tool lets users do. You won't get an understanding for Fitzgerald or Stoker's unique styles, or even their plots, using AI. But you can make two characters kiss. 

I didn't feel that I was embodying Nick. I felt like I was temping as a fan fiction author. Albeit, an author without full control of my own story and with main characters who might not fully understand their own world. 

AI is rewriting fandom rules

Fan fiction is dominating publishing trends and moving Hollywood money. Companies are leaning into the more taboo parts of fandom, like erotica, to profit from the mainstreaming of fan culture. Generative AI itself has been positioned by companies as a boon to fandom.

Meanwhile, those actually in fandoms are failing to connect AI with their worlds.

AI skeptics (I am admittedly one) worry that the tech is a threat to the core tenants of fandom: human creativity and connection. As companies advertise AI as a creative tool to strengthen fandoms, specific fandom practices are more at risk than others, particularly the parts of it that are generative and novel, like fan art, zines, and, primarily, fan fiction. 

Now via AI tools like Books, users can role play as a character without even engaging with the text itself. The immediacy is alluring. You can easily forget the satisfaction of writing a "Fix It" fic for your fandom friends or scouring through tags of human-made stories for the perfect AU. But all of that, I promise, is better than the bot. And unlike online fan spaces, where profit is a faux pas, you may have to cough up cash to do it.

Through clever marketing and the allure of AI itself, particularly the ease and immediacy of it, Character.AI is repackaging the decades-old tradition of fan fiction and selling it back to you. The platform's shiny new features — alternate universes, canon divergence, original characters — are the bones and sinew of fan works. Fanfiction.net, Archive of Our Own (AO3), and Wattpad writers, many of whom are now published authors, would take one look at this and turn the other way. That's what we do, they'd scoff.

This article reflects the opinions of the writer.

Have you used an Android phone in the past nine years? Then Google might have to give you up to $100 later this year.

That's because the company reached a preliminary $135 million settlement (without admitting wrongdoing) in a class-action lawsuit called Taylor v. Google LLC, per CNET. The suit alleged that Google used Android users' paid cellular data to transfer information to Google without their permission. Now, users who may have had their data misused can sign up for payments on the official settlement website.

The settlement could include up to 100 million Android mobile users in the United States. If you think you qualify, check the email associated with your Android mobile account for the settlement notice.

Not sure if you're eligible? Here are the criteria:

1. You have to be a real human in the United States

2. You have to have used an Android phone with cellular data at any point between Nov. 12, 2017 and now

3. You can't be a member of Csupo v. Google LLC, a similar class-action lawsuit specifically for Californians

SEE ALSO: Android adds custom caller ID cards, new location sharing features

If you meet those requirements (and surely a whole lot of people do), you can enter your payment information on the settlement website. There's a final hearing on June 23 to determine whether or not these payments will actually go out, so you'll know by then if you're getting any money. And no, we don't know exactly how much each affected user will get, though payments are capped at $100. That doesn't mean anyone will actually get $100, though.

While Google has, again, not admitted wrongdoing here, it has agreed to pay out the settlement and will also update its Google Play terms of service regarding passive data transfers using cellular data. But really, the thing that matters here is that you might get a little bit of walking-around money for something you didn't even realize happened several years ago.

Heard lots of buzz about Netflix's Something Very Bad Is Going to Happen, but need to know more about the story before taking the plunge and checking it out? Don't worry, we've got you covered.

SEE ALSO: 'Something Very Bad Is Going to Happen' review: Marriage is a killer

The eight-episode miniseries, created by Haley Z. Boston and executive produced by Stranger Things creators the Duffer Brothers, is a non-stop rollercoaster of wedding anxiety. It kicks off with the introduction of engaged couple Rachel Harkin (Camila Morrone) and Nicky Cunningham (Adam DiMarco), who are off to the woods for an intimate ceremony at Nicky's family's cabin.

Has Rachel ever been to this cabin? No. Has she even met Nicky's family? Also no. The red flags are piling up, and making matters worse is the fact that Rachel has an unshakeable feeling that, well... Something very bad is going to happen.

What could that something very bad be? Is it related to Nicky's odd family, who are acting extra shady about the wedding? Is it tied to a Cunningham horror story about the bloodthirsty Sorry Man who lurks in the woods? Or might it have to do with the strange man (Zlatko Burić) who keeps following her around, asking if she's sure Nicky is the one?

All these questions and more combine to form Something Very Bad Is Going to Happen's atmospheric puzzle, one that dives deep into the perils of wedding anxiety and stress about finding your true soulmate. Press play for a spooky binge that doubles as one of Netflix's best 2026 offerings so far.

Something Very Bad Is Going to Happen is now streaming on Netflix.

Want to see Fitbit's new Whoop-style screenless fitness tracker? Start watching Golden State Warriors games.

Or, at the very least, start stalking Warriors star and NBA legend Steph Curry's Instagram. Curry recently teased the new fitness tracker in a sponsored Instagram post with Google, which owns Fitbit. Along with short clips of Curry wearing the tracker, a caption reads, "#sponsored I won’t spoil it. You kinda have to see it for yourself 👀"

As reported by Droid Life, Curry has been seen wearing the mysterious wrist-worn fitness tracker in public for the past several months. On April 15, photographers spotted Curry wearing the device before a game against the LA Clippers at the Intuit Dome in Los Angeles, California, giving us an up-close, high-resolution look at the grey-and-orange fitness tracker.

Left: Move the slider to zoom in on the device. Credit: Adam Pantozzi/NBAE via Getty Images Right: Credit: Adam Pantozzi/NBAE via Getty Images

The unnamed mystery device Curry is teasing has not been officially announced, but we know it's a screenless fitness tracker from Google.

Bloomberg reported on March 31 that Google was developing exactly this type of screen-free fitness band for Fitbit, and that Curry was involved. That Bloomberg report also said the device would come with a paid subscription for extra features. And, in the most predictable development possible, the device will also come with AI features, specifically an "AI-powered Fitbit personal health coach" available in the Fitbit app.

SEE ALSO: Google wants to fill Fitbit with AI — and your medical records

This device looks a lot like the popular Whoop fitness tracker, which Mashable has put to extreme real-world testing in the past.

Unlike something like the Google Pixel Watch 4 or a traditional Fitbit, there's no screen, meaning you spend less time looking at it. Crucially, its battery also lasts a heck of a lot longer, making it an ideal sleep tracker as well. As for this mystery Fitbit device, it looks slightly thinner than Whoop's hardware.

Bloomberg didn't have any specifics on a possible launch date beyond "later this year." However, since Curry has been wearing one in basically every public appearance he has made in the last few months, we wouldn't be surprised if the device launched sooner rather than later.

Anthropic released a new hybrid reasoning model on Thursday: Claude Opus 4.7.

Anthropic has a reputation as a safety-first AI company, and the Opus 4.7 system card reports that the model is less likely to hallucinate or engage in sycophancy than both prior Anthropic models and other frontier AI models.

We dived into the Opus 4.7 system card to see exactly what Anthropic had to say about the model's safety, honesty, and sycophancy.

Don’t miss out on our latest stories: Add Mashable as a trusted news source in Google.

The TL;DR version

Why put the TL;DR version at the end?

Anthropic says Claude Opus 4.7 makes improvements on various types of hallucinations and overall honesty. Anthropic gives the new model top marks on sycophancy and encouragement of user delusions, too. (Anthropic's data also shows that Claude Opus 4.7 scores much better on these behaviors than Gemini 3.1 Pro and Grok 4.20.)

"Claude Opus 4.7 is more reliably honest than Opus 4.6 or Sonnet 4.6, with large reductions in the rate of important omissions, and moderate improvements in factuality and rates of hallucinated input," Anthropic reports.

False premises honesty rate: Will the model tell a user when they're incorrect? Credit: Anthropic MASK honesty rate: Will the model contradict its own stated belief when pushed to do so by a user? Credit: Anthropic

Want to learn more about getting the best out of your tech? Sign up for Mashable's Top Stories and Deals newsletters today.

Anthropic measures Claude's honesty and hallucination rates in multiple ways, but let's look at one representative example — the Model Alignment between Statements and Knowledge (MASK) benchmark. MASK was developed by Scale AI and the Center for AI Safety.

Claude Opus had a MASK honesty rate of 91.7 percent, compared to 90.3 percent for Opus 4.6 and 89.1 percent for Sonnet 4.6. While that’s lower than the 95.4 percent score achieved by Claude Opus 4.5, the new model performs better on other hallucination scores (more on that below).

Interestingly, Claude Mythos was more honest still, with an honesty rate of 95.4 percent.

Claude Opus 4.7 lags behind Claude Mythos on overall performance

Since Anthropic repeatedly compares Opus 4.7 to Claude Mythos, let's quickly review the differences between the two models.

Claude Opus 4.7 is the latest hybrid reasoning model available to paid Claude subscribers. Claude Mythos is an unreleased model that Anthropic has only made available to partners via Project Glasswing.

SEE ALSO: Anthropic makes the case for anthropomorphizing AI in ‘unsettling’ research paper

Under normal circumstances, we would expect Claude Opus 4.7 to be Anthropic's most advanced and powerful model to date. However, Anthropic says it lags behind the unreleased Claude Mythos in key areas. Anthropic deemed Claude Mythos too dangerous to release to the public because of its advanced cybersecurity capabilities.

Still, Claude Opus 4.7 improves upon Opus 4.6 in many ways, particularly advanced coding, visual intelligence, and document analysis, Anthropic says.

More details on Claude Opus 4.7 hallucination rates

When using Opus 4.7, how likely is Claude to tell a lie, invent facts, or deceive users? There isn't a single hallucination rate that Anthropic provides, because there are multiple types of hallucinations.

So, this section is for the AI nerds.

Anthropic identifies a few different ways to measure hallucination and honesty:

• Factual hallucinations: How likely the model is to provide accurate information. How often does the model admit that it doesn't know something?

• Input hallucination: This occurs when an AI model ignores prompt instructions, hallucinates the content of files, or pretends to have access to a tool it doesn't have.

• False premises honesty rate: Will the model tell a user when they're incorrect?

• MASK honesty rate: This "tests whether a model will contradict its own stated belief when a user or system prompt pushes it to."

We've already covered the MASK honesty rate, and Claude Opus 4.7 shows similar gains on these other measures, according to Anthropic.

At this time, we cannot independently verify Anthropic's results.

To measure factual hallucinations, Anthropic used four different tests and recorded correct responses, incorrect responses, and abstentions. In this case, abstentions are good — the model should decline to answer a question rather than guessing. Across all four tests, Opus 4.7 scored higher than Opus 4.6 and Sonnet 4.6 but lower than Claude Mythos.

Chart showing Claude Opus 4.7's performance on accuracy tests. Credit: Anthropic

Anthropic measured Opus 4.7's input hallucination in two ways: "prompts requesting an unavailable tool" and "prompts referencing missing context."

Opus 4.7 scored 89.5 percent on the former, beating Claude Mythos's 84.8 percent; on the latter, Opus 4.7 scored 91.8 percent, two points lower than Claude Mythos's 93.8 percent.

This shows just how stubborn AI hallucinations are, with even leading AI companies like Anthropic recording input hallucination rates around 90 percent. Anthropic's reported hallucination rates are similar to the latest OpenAI models, which provide responses with incorrect information up to 5.8 percent of the time (with browsing enabled) to 10.9 percent (browsing disabled), per OpenAI.

OpenAI most recently reported hallucination rates in the system card for GPT-5-2. Credit: OpenAI

What about Opus 4.7's honesty rate for false premises, i.e., will Claude tell a user they're wrong? According to the system card, Claude will push back on false premises 77.2 percent of the time. That's better than all other recent Anthropic models except for — you guessed it — Claude Mythos, which will reject false premises 80 percent of the time.

SEE ALSO: Google AI overviews: Confident when wrong, yet more visible than ever Claude Opus 4.7 sycophancy

There's not much new to report in terms of sycophancy. While Anthropic's expert red-team testers reported that Opus 4.7 was prone to “sycophantic agreement under pushback," it has very similar scores to prior models from Anthropic and OpenAI, and noticeably better scores than Gemini 3.1 Pro and Grok 4.20. Again, this is according to Anthropic.

To measure bad behaviors like sycophancy and "encouragement of user delusion," Anthropic uses Petri 2.0, its open-source behavioral audit tool. This test scores models on a 1-10 scale, with lower scores reflecting better behavior. The Petri score isn't akin to a percentage, as it measures both the rate of a behavior and the severity.

Anthropic scored Opus 4.7 highly (or, lowly, with this particular scale) on both sycophancy and user delusions.

Anthropic uses Petri 2.0, its open source AI safety tool, which scores bad behaviors from 1-10. The lower the score, the better. Credit: Anthropic

Mashable reached out to Anthropic for comment but did not receive a response in time for publication.

Disclosure: Ziff Davis, Mashable’s parent company, in April 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.

Netflix's horror series Something Very Bad Is Going to Happen ended, as its title suggests, with something very bad: a wedding bloodbath that left almost all the guests dead.

SEE ALSO: 'Something Very Bad Is Going to Happen's bloody ending, explained

However, it also ended with the promise of something good: a fresh start for heroine Rachel Harkin (Camila Morrone), who survived a family curse and escaped a marriage with the spineless Nicky Cunningham (Adam DiMarco). The curse that plagued the Harkin family for generations has since passed on to the Cunninghams, and the now-immortal Rachel will act as the witness for any of their future weddings. Here's hoping she has a much less frightening aura than the witness who haunted her family, played by the formidable Zlatko Burić.

By the finale's end, Rachel's arc seems fairly complete. But is there a chance Something Very Bad Is Going to Happen will come back for a Season 2?

Has Netflix renewed Something Very Bad Is Going to Happen for Season 2? Camila Morrone in "Something Very Bad Is Going to Happen." Credit: Netflix

So far, no. Netflix has not indicated any plans for a second season of Something Very Bad Is Going to Happen. Part of that is by the show's design: It was originally envisioned as a limited series. However, several miniseries have grown so popular they've picked up second seasons. Look at Big Little Lies, or The White Lotus, or even Netflix's own Beef, now an anthology. Could Something Very Bad Is Going to Happen follow a similar route?

In an interview with TheWrap, series creator Haley Z. Boston noted that the possibility for a Season 2 is there, but that it would look fairly different.

"There is an open thread, but this was so inspired by my own fear that I'm gonna need another existential fear to explore," Boston told TheWrap. "I think we're done with the wedding thing."

Something Very Bad Is Going to Happen is now streaming on Netflix.

Horror series Something Very Bad Is Going to Happen delivers atmospheric thrills and chills by the bucketful, so if you haven't already checked it out, consider this your sign to add it to your watchlist.

SEE ALSO: 'Something Very Bad Is Going to Happen' review: Marriage is a killer

Created by Haley Z. Boston (Brand New Cherry Flavor) and executive produced by Stranger Things creators the Duffer Brothers, the eight-episode miniseries is now streaming on Netflix. It premiered March 26 and pulled in 28.3 million hours viewed in its first week, according to Netflix. It's remained in Netflix's top 10 series globally every week since its release, although as of this writing, it no longer appears on Netflix's top 10 TV shows carousel.

The series follows engaged couple Rachel Harkin (Camila Morrone) and Nicky Cunningham (Adam DiMarco) in the week leading up to their wedding at Nicky's parents' remote cabin. As their special day approaches, Rachel can't shake the feeling that — you guessed it — something very bad is going to happen. (Maybe the remote cabin should have been the first red flag.)

Based on Nicky's off-putting family, you may expect that "something very bad" to look like something out of Ready or Not, where a filthy rich family turns on their newest addition. But Something Very Bad Is Going to Happen takes a different turn, looping in strange curses and family trauma to examine anxieties around marriage and finding the perfect soulmate.

The carnage that follows certainly lives up to the show's title, creating one of the most engrossing Netflix binges of 2026.

Something Very Bad Is Going to Happen is now streaming on Netflix.

Bluesky, the social media app popular with X expatriates, suffered a widespread outage on Thursday, April 16.

And in a thread posted on the official Bluesky profile, the app's leaders revealed the cause of the outage — a distributed denial-of-service (DDoS) attack.

"Our team received a report of intermittent app outages at about 11:40pm PDT on April 15, 2026," the post read. "They worked through the night to mitigate a sophisticated Distributed Denial-of-Service (DDoS) attack, which intensified throughout the day."

Fortunately, Bluesky also said there was no evidence that any user data had been compromised in the attack, which affected the Bluesky app, feeds, notifications, and search. The company said it would provide further information by 1 p.m. ET Friday.

Our team received a report of intermittent app outages at about 11:40pm PDT on April 15, 2026. They worked through the night to mitigate a sophisticated Distributed Denial-of-Service (DDoS) attack, which intensified throughout the day.

— Bluesky (@bsky.app) April 16, 2026 at 7:47 PM

The Bluesky status page indicated the app was fully online as of this writing; the app reports a 99.983 percent uptime over the past 90 days.

A DDoS attack is relatively simple and low-effort for cybercriminals. In this type of attack, hackers send a massive number of requests to overwhelm servers. DDoS attacks have been around since the early days of the World Wide Web, and as the Bluesky outage shows, they can still cause problems.

The Bluesky outage began affecting users in the early morning hours, with the service DownDetector recording thousands of user error reports. (Disclosure: DownDetector and Mashable are both owned by the same parent company, Ziff Davis.)

Bluesky saw massive growth after Elon Musk took over Twitter, and again following the second election of President Donald Trump. However, its growth has stalled more recently, and data shows that daily active users on Bluesky have declined.

Mashable reached out to Bluesky with questions about the attack, and we'll update this story if we receive a response.

We're still several months away from Apple showing us the iPhone 18, but we can still bask in leaks until then.

This week, we've got a pair of leaks involving the cameras and colors for the upcoming flagship iPhone.

On the camera side, Korean outlet ETNews (via MacRumors) reports that Apple will debut a new variable-aperture rear camera system on the iPhone 18 Pro and Pro Max in September. Prior to this year, all iPhones had fixed-aperture cameras, meaning the lens remained completely open during image capture. A variable aperture will allow the lens to open and close gradually depending on the situation, letting in more light in darkness or keeping excess light out in bright settings.

SEE ALSO: Apple iPhone 17e review: Ticks every box but one

That would give iPhones an even greater level of flexibility in photography, and this report is backed up by prominent Apple analyst Ming-Chi Kuo, who said Apple would incorporate a variable aperture in the iPhone 18 all the way back in 2024.

What do the latest iPhone rumors have to say about colors?

Macworld claims to have received some information about Apple's chosen colors for 2026. According to their sources, the headline color this year will be "Dark Cherry," a dark red that will replace the controversial cosmic orange color from the iPhone 17 Pro. Macworld also said that light blue, dark gray, and silver options are also possibilities for the iPhone 18 Pro. A previous leak also suggested Apple would ditch black for the iPhone 18 Pro, which may not go over well if it proves true.

All will be revealed in September, though the iPhone Fold may get most of the attention.

TL;DR: The MacBook Neo is sold out at Apple (shipping dates have slipped into May), but you can still buy right now for Amazon's best-ever price of $589.99.

Opens in a new window Credit: Apple MacBook Neo (256GB) $589.99 at Amazon
  Shop Now

If you thought Apple’s MacBook Neo was just a niche product for students, think again. The demand for Apple's latest entry-level laptop has been staggering. Don't believe us? Shipping dates have slipped into May if you order direct from Apple.

That's the bad news. The good news is that whether you want the 256GB base model or the upgraded 512GB version, Amazon has you covered. And better yet, you can order right now for Amazon's best-ever price of $589.99.

SEE ALSO: Testing the MacBook Neo on 8K video and raw photos

So how did we get here? Global RAM shortages have jacked up the prices of Windows laptops, making the sub-$600 MacBook Neo with the A18 Pro chip the most logical purchase on the market. In Mashable's review of the MacBook Neo, Timothy Werth sums it up: "For students, young people, and casual users, there's nothing else like it at this price."

We're not rushing you into any decision here, but if you were already contemplating a purchase, consider this a warning. The popularity of the MacBook Neo is surging, and Amazon's record-low price might not stick around forever.

Apple Watch Series 11, Garmin Venu 4, Google Pixel Watch 4, Samsung Galaxy Watch 8, and Amazfit Bip 6 were tested over a 30-mile run. The comparison focused on real-world tracking accuracy across distance and performance metrics. One smart watch ultimately delivered the most consistent results across the test.

The calculus for buying a new Mac has officially changed. With the debut of the $599 MacBook Neo and a massive power bump for the M5 series of processors, Apple's 2026 laptop lineup looks very different from what it did just months ago.

In this Lab Report, we’re breaking down Apple’s entire MacBook stack—from the new budget Neo to the monster M5 Max MacBook Pro—to help you decide which machine actually belongs on your desk. We’ve spent hours testing these in the PC Labs to see how each type of MacBook handles real-world workflows.

SAVE 46%: As of April 17, you can get the Monster Shock portable Bluetooth speaker for just $26.99, down from $49.99. That's a 46% discount or a $23 price cut.

Monster Shock portable Bluetooth speaker $26.99 at Amazon
$49.99 Save $23.00   Get Deal at Amazon

Summer's right around the corner (for some of us, it feels like it's already here!), and that means bonfires, beach days, and backyard BBQs, which obviously, couldn't be had without some tunes.

If your speaker is shot or you've yet to get a portable Bluetooth speaker, the Monster Shock portable Bluetooth speaker is on sale for $26.99 at Amazon, down from $49.99. That's the lowest price we've tracked for this speaker, but Amazon's marking it as a "limited time deal," so you'll need to act fast.

SEE ALSO: I wore JLab's ridiculously big 'headphone' speakers. I still have so many questions!

The Monster Shock is IPX6 waterproof (so you can take it to the beach or bring it to the pool), has a 30W power driver and a max Bluetooth range of 100 feet, and offers 17 hours of playtime. And because it's practically half off right now, you could even grab two and pair them together for True Wireless Stereo surround sound for just a few bucks more than the original price of one.

Before you buy: a few reviewers noted that the speaker's automated control voice is aggressively loud, so maybe don't power it on for the first time while your friends are trying to nap in the sun.

DJI has unveiled the Osmo Pocket 4, the next iteration of its popular handheld gimbal camera. Sadly, it may be a while before people in the U.S. get to try it out.

Since its release in 2023, DJI's Osmo Pocket 3 has become a stalwart favorite of content creators across the globe. The newly announced Osmo Pocket 4 now adds several upgrades to the familiar 4K vlogging camera, such as dedicated zoom and customizable buttons that are accessible when the 2-inch touchscreen is twisted into landscape mode.

Want to learn more about getting the best out of your tech? Sign up for Mashable's Top Stories and Deals newsletters today.

SEE ALSO: I reviewed the DJI Osmo Pocket 3 and I can't imagine a better social video camera

The DJI Osmo Pocket 4 camera features a 1-inch CMOS sensor, with 14 stops of dynamic range, helping it deliver improved low-light performance. Capable of filming at a maximum resolution of 16:9 4K at 60fps, the Osmo Pocket 4 can also record 4K slow-mo footage at 240fps — a frame rate the Pocket 3 only reaches at 1080p. 

DJI states that the Pocket 4 can run for an impressive four hours at its lowest resolution of 1080p 24fps, which is notably longer than the Pocket 3, and also has 107GB of built-in storage. This will likely be welcome news to anyone concerned about the increasing cost of memory cards, as the Pocket 3 relies entirely upon SD cards.

Credit: DJI The DJI Osmo Pocket 4 camera is not yet approved for U.S. release

Unfortunately, a U.S. release of the DJI Osmo Pocket 4 camera has not yet been confirmed. While other parts of the world will get their hands on the Osmo Pocket 4 within the next week, the U.S. government's hostility toward DJI continues to significantly hamper its business in the country.

SEE ALSO: We talked to DJI about the FCC drone ban in the United States.

A DJI spokesperson told Mashable that it has requested authorization from the FCC to release the Osmo Pocket 4 in the U.S. market; however, its application is still pending. And given the FCC's tough stance on foreign drone companies, U.S. content creators will simply have to continue waiting.

Meanwhile, in Australia, pre-orders for the DJI Osmo Pocket 4 are open now, with shipping scheduled to begin on April 22. The Standard Combo bundle costs $AU769 (approximately $551) and includes a handle, carrying pouch, gimbal clamp, and wrist strap. The $AU959 (approximately $687) Creator Combo includes all that plus a wide-angle lens, fill light, mini tripod, carrying bag, and a DJI Mic 3 transmitter with accessories.

Credit: DJI

At first glance, many may have assumed that Bob Odenkirk's new film, Normal, was another sequel in the Nobody franchise. With the same star, writer, and producer, the two films do share much of the same DNA. However, the team's new movie is not only a completely new story but attempts to blend genres in a way that further separates it from its predecessor.

Mashable's Entertainment Editor, Kristy Puchko, caught up with the Normal team at SXSW to talk about the film, its themes, and why they chose to film on 35mm.

The film's writer, Derek Kolstad (John Wick, Nobody), noted how this film differs from Nobody, the team's previous collaboration. "To this point, you know, in Nobody, if there's a house on fire, Hutch is going to run into it and hopefully fight anyone who's in there. In Normal, he's going to look at the house on fire and say, 'I don't know.'"

Odenkirk added, "It's not that much fire." The Normal star described the film as "a story of a guy who has lost trust in himself, his instincts. In the course of interacting with the people of this small town who are hiding something from him, he is desperately trying to avoid seeing it. And then his hand is forced. Once he sees it, he has to re-engage with the situation and the world around him. And then there's a flip-floppery of allegiances that is very surprising and probably a bit inscrutable. When asked to define, sort of, the subtextual themes, I'm like, 'Wow, there's like six of them,' and they're a little bit conflicting. And that's what's great about it."

Ben Wheatley (The Meg, Kill List, Rebecca), the latest addition to the production team, when asked what attracted him to the film, replied, "In the kind of range of things I still want to make. It basically has the DNA of a cowboy movie, and it has it has the white hat, black hat kind of style. It has a sheriff, which is a point of morality in the middle. But it still speaks to my kind of slightly twisted kind of sense of morality as well."

In the full interview on YouTube, the team also talks about the undeniable difference of filming with practical effects as opposed to CGI, the way certain movies hit differently as we get older, and why it was important that Normal be an homage to the classic action-thriller genre. "I think one of the reasons we're projecting this on 35 millimeter is because we want to point to that quality," Odenkirk said. "There's a satisfaction, I hope, that we're achieving. That was the reason for the poster that we have. The poster is really a throwback visual."

Normal is now in theaters.

JD Christison walks through how to set up Starlink Roam, a satellite internet service designed for travelers. Follow along as he conducts a speed test and a work call in the middle of nowhere, while sharing essential tips and tricks of the network.