Claude Opus 4.7

Welcome to the Tech Brew right home for Thursday, April 16th, 2026. I'm Brian McCullough. Today, Anthropic dropped Claude Opus 4.7 for coding and rolled-out government ID verification for some users. TSMC beat earnings and raised forecasts on strong AI demand. X launched cash tags and Google SpaceX. Here's what you missed today in the world of TV.

Anthropic has released Claude Opus four point seven saying it is a quote notable improvement on Opus four point six in advanced software engineering and comes with a new X high effort level. Now, to be clear, Anthropic also said Opus 4.7 is less broadly capable than Claude Mythos Preview, and that its cyber capabilities are not as advanced as those of Mythos Preview, but You can't use Mythos preview right now, so that's neither here nor there.

Quoting Bloomberg. Anthropic said Thursday that Opus four point seven is meant to be better at software engineering, including fielding some of the hardest coding tasks that previously required greater supervision. However, the new model is less broadly capable than Mythos, including for cybersecurity uses, Anthropics said.

During the training process for Opus four point seven, Anthropic went so far as to experiment with ways to differentially reduce the model's cyber capabilities, according to a company blog post. Last week, Anthropic warned that its Mythos system was able to identify and then exploit vulnerabilities in every major operating system and every major web browser when directed by a user to do so.

The company decided to only make a version of the model available to select businesses to help them safeguard their software. We are releasing OPIS four point seven with safeguards that automatically detect and block requests that indicate prohibited or high-risk cybersecurity uses, the company said in the blog post.

What we learn from the real-world deployment of these safeguards will help us work towards our eventual goal of broad release of Mythos class models, end quote. And quoting nine to five Mac. Quoting anthropic.

Opus four point seven is a notable improvement on Opus four point six in advanced software engineering with particular gains on the most difficult tasks. Users report being able to hand off their hardest coding work, the kind that previously needed close supervision to Opus four point seven with confidence. Opus 4.7 handles complex long running tasks with rigor and consistency, pays precise attention to instructions, and devises ways to verify its own outputs before reporting back.

Anthropic says its model has better vision and more taste for creating higher quality work. The model also has substantially better vision. It can see images in greater resolution. It's more tasteful and creative when completing professional tasks, producing higher quality interfaces, slides, and docs.

The company shows favorable benchmarks across a range of uses, including agentic coding and computer use that put Opus four point seven ahead of four point six, GPT five point four, and Gemini three point one pro, but behind the more broadly capable Cloud Mythos preview. Anthropic highlights improvements to instruction filing, multimodal support, real-world work, and memory, as other improvements in Opus 4.7.

Opus four point seven is better at using file system based memory, the company says. It remembers important notes across long multi-session work and uses them to move on to new tasks that, as a result, need less upfront context. Notably, Anthropic has established a more predictable cadence for directly upgrading its Clawed Opus model. Opus 4.7 arrives two months after Opus 4.6, which arrived two months after Opus 4.5.

There was a three-month gap between 4.1 and 4.5. And Thrapic's announcement includes a note to users on how token usage is handled with Opus 4.7 quote. Opus four point seven is a direct upgrade to Opus four point six, but two changes are worth planning for because they affect token usage. First, Opus four point seven uses an updated tokenizer that improves how the model processes text. The trade off is that the same input can map to more tokens, roughly 1 to 1.35x, depending on the context type.

Second, Opus thinks more at higher effort levels, particularly on later turns in agentic settings. This improves its reliability on hard problems, but it does mean it produces more output tokens. Starting today, auto mode and cloud code is also available for max plan subscribers, not just Teams slash enterprise slash API customers.

Claude Code also gains the new slash ultra review command that Anthropic says runs a dedicated review session that reads through your changes and flags what a careful reviewer would catch. End quote.

TSMC reported Q1 revenue up thirty-five point one percent year on year, net income up fifty-eight point three percent year on year, both above estimates, blah di blah de blah. But what is more interesting for our purposes is that T S M C C E O C C Y Said TSMC checked with customers about AI demand and was reassured that it was still strong even amid the Iran War, and it is raising revenue forecasts accordingly. Quoting the journal.

TSMC, a contract chipmaker for clients such as NVIDIA and Apple, projected that its capital expenditures for the year would hit the higher end of the fifty-two to fifty-six billion dollar range projected in January. The company said it expected revenue to grow more than thirty percent this year in dollar terms from an earlier estimate of around thirty percent.

TSMC's chief executive, CCY, said the company had double-checked with customers about AI demand, and they reassured the chipmaker that it was still strong. That led TSMC to accelerate investments, such as building more clean rooms, he said. The company reported a gross profit margin of sixty-six percent for the first quarter, the highest level in more than twenty years, according to data from SP Global Market Intelligence.

Despite weaknesses in some segments, including an eleven percent decline in its smartphone business compared with the previous quarter. For anyone wondering whether the AI trade still has legs, TSMC just told us that business has never been better, Josh Gilbert, an analyst at eToro said in a note.

TSMC's aggressive capital spending and lofty profit margins have made it the nerve center of the global semiconductor industry. Equipment and material makers are clustering around its expanding hubs in Taiwan, the US, and beyond. They are aligning with TSMC's roadmap to get their share of the profit, further cementing TSMC's leadership in the global chip supply chain.

Taiwan's benchmark stock index has notched record highs in the past few sessions, including on Thursday when it was up one point one percent. TSMC's shares closed to a new high Thursday, giving the company a market capitalization of one point seven trillion after an initial drop following the outbreak of the Iran War in late February.

Taiwan has other AI-related suppliers such as Foxconn, which builds NVIDIA's servers and companies that specialize in the technology to stack and integrate chip components, an essential element in building AI processors.

The Middle East conflict has done little to damp the optimism, but it still poses risks. Taiwan relies heavily on imported fuel for electricity, and chip production is power intensive. Last week, TSMC Vice President Cliff Howe Called on Taiwan to increase strategic reserves and diversify procurement channels of helium, hydrogen, and natural gas to ensure a stable supply.

TSMC executives said Thursday that with Taiwan securing sufficient liquefied natural gas supplies through at least May, the company doesn't expect near term disruptions to production. They said specialty gases such as helium and hydrogen were sourced from multiple suppliers and regions, and any issues were likely to have a material impact on chip output. Qatar supplies a chunk of the world's helium, which is needed for cooling in chip production. End quote.

Turn your market hunches into trades with liquid. Liquid lets you go long or short on commodities, stocks, or private markets right from your phone with up to 100x leverage. That means a$10 position can give you up to$1,000 in market exposure. And you don't have to work around market hours or weekends with Liquid. You get one platform for every market open twenty-four-seven. More than three billion dollars has already been traded on Liquid. Check them out at liquid.trade.com.

Slash TechBrew and sign up with referral code TechBrew for a$25 deposit match. Terms apply. Some days it just feels like my entire job is keeping up with the paperwork. I've got, you know, forms for taxes right now, contracts scheduling for, you know, interviews. Sucking up every single spare minute I have. That's why upgrading to

Gusto is such a game changer. Gusto is online payroll and benefit software built for small businesses. It's all-in-one, remote friendly, and incredibly easy to use, so you can pay hire on board and support your team from anywhere. Gusov's got it all. Payroll tax filing, direct deposits, health benefits, commuter benefits. Workers comp even four one case. They've even got tons of options for pretty much all budget sizes.

Try gusto today at gusto.com slash brew and get three months free when you run your first payroll. That's three months of free payroll at gusto.comslash brew.

Back to Anthropic though, cause this is interesting. Anthropic has quietly rolled out identity verification that may require clawed users to provide a government-issued photo ID and live selfie to access certain capabilities. Quoting to Cripp. We are rolling out identity verification for a few use cases, and you might see a verification prompt when accessing certain capabilities as part of our routine platform integrity checks or other safety and compliance measures, Anthropic said.

We only use your verification data to confirm who you are and not for any other purposes. Millions of users fled OpenAI for Anthropic in February after OpenAI signed a deal to deploy AI on Pentagon classified networks, a contract Anthropic turned down over concerns about mass surveillance and autonomous weapons. Daily signups broke records and free users were up sixty percent since January, Anthropics said at the time. Their privacy conscious crowd had found its home.

That crowd, it seems, may now have some documents to prepare if it wants to continue using Claude. The reactions so far have been quite negative, pointing out that it's a deliberate decision and not a regulation or a mandatory order imposed by a government on anthropic as a service provider. Quoting Ryan Sean Adams on X. AI know your customer is here. New cloud subscribers ask for government ID and photo. Not even a regulatory requirement. Anthropic just doing it because they want to.

But regulatory is coming. Next up will be laws. No ID without government issued ID. All AI use tracked to individuals. No private AI. End quote. According to the Help Center page, which went live on April 14th, Anthropic selected persona identities as its verification partner, the same know-your customer infrastructure used across financial services.

and requires a physical, undamaged passport, driver's license, or national identity card. Photocopies, mobile IDs, and student credentials don't count. A live selfie may also be required. The policy isn't universal yet and may not ever be. Verification will trigger when accessing, quote, certain capabilities during routine platform integrity checks or as part of safety and compliance measures.

Anthropic had not commented publicly on the policy by the time users raised concerns on social media, but a spokesperson told decrypt this applies to a small number of cases where we see activity that indicates potentially fraudulent or abusive behavior, which is against our usage policy. On data handling, Anthropic draws a careful line. Your ID and selfie go to Persona's servers, not Anthropic's own system.

The company says it is the data controller setting the terms and that persona can use the information to verify identity and improve fraud detection. The data is encrypted in transit and at rest, excluded from model training and won't be shared with third parties for marketing, something Anthropic has been careful to promise since its earliest commercial policies. Careful promises though have a history of meeting careless infrastructure.

In October 2025, breach at Discord exposed roughly 70,000 government IDs users had submitted for age verification. Persona is a serious player in the space, but third party custody of government documents has demonstrated repeatedly that no third party is immune. Tighter identity controls also fit a pattern Anthropic has been building toward. In December, the company announced classifiers to detect users who self-identify as miners.

Multiple adult users had their accounts suspended anyway, reporting that entire project histories were wiped while they tried to appeal incorrect flags. Accounts registered from regions Anthropic doesn't formally serve are also subject to bans, a detail that lands hardest on Chinese users accessing Claude through intermediaries, since a live selfie matched against a physical government document is hard to fake your way through. End quote.

X has officially launched cash tags, something that has informally existed for years, a feature that shows real-time financial data on stocks and crypto directly in users' timelines, starting with iOS in the US and Canada. quoting the block. When a user searches for a specific asset or a contract address, X will automatically suggest related stocks or cryptocurrencies as a cash tag. Users can also tap a cash tag to browse posts mentioning that asset alongside a live price chart.

Canadian users will see a built-in button linking to Wealth Simple, a Canadian brokerage to trade assets from the app. This ensures that you're always matched to the chatter for the right stock or tokens, said Nikita Beer, head of product at the social media platform. Cashtags are just the first step in our commitment to be the best destination for the finance and crypto community. It's not immediately clear whether or when the feature will be expanded to other regions.

The block has reached out to X for further information, but has not yet heard back. The launch came a day after Beer noted on X that crypto has had a rough year. Maybe we should launch something to fix it, he said. Beer also previously clarified that X will not execute trades or act as a brokerage, but intends to build financial data tools and links.

The cash tag's feature aligns with Elon Musk's broader ambition to transform X into a super app. Musk said last month that the company plans to launch early public access to X Money in April, a digital wallet designed to enable peer-to-peer transfers and payments. A December job posting showed that X was recruiting a senior engineer to help build the X Money payments platform, though it did not specifically reference cryptocurrency or blockchain.

Last month, X hired Benji Taylor, a former design lead at Ave, and BACE as its head of design.

Finally today, Google is about to make some serious bank, like serious bank beyond the serious bank they already make, seriously. Quoting Bloomberg An early investment in SpaceX has positioned Alphabet for a twelve figure windfall, a new filing shows, underscoring the vast wealth likely to be created by the rocket company's market debut.

Google LLC owned a six point one one percent stake in Elon Musk's company at the end of twenty twenty five, according to a new disclosure the space startup filed this week in Alaska, where firms must report holders with stakes of five percent or more. At a$2 trillion valuation, which SpaceX hopes to exceed in its initial public offering, a holding of that size would be worth$122 billion.

Google's stake has likely been diluted following the February merger of SpaceX with XAI, Musk's artificial intelligence and social media company. It now likely owns roughly five percent of SpaceX following the transaction according to Bloomberg calculations, which would be worth a hundred billion dollars at a two trillion dollar IPO valuation.

Google and Musk himself, with a roughly forty percent stake, were the only entities required to disclose their holdings in the filing, but several other individuals and firms are also poised to make billions off the listing. SpaceX has filed confidentially to go public, targeting a June listing, and is expected to raise as much as$75 billion in what would make it the biggest ever IPO, Bloomberg previously reported.

If it achieves a market valuation of$2 trillion, a 0.05% stake could turn a shareholder into a billionaire overnight. The listing would also cement Musk as the world's first trillionaire and lift the personal fortunes of longtime lieutenants, including President Gwen Shotwell. Its earliest backers will likely see huge returns on their money, but even those who got in five years ago will still be happy with the outcome, said Pitchbook senior research analyst Franco Granda, who covers SpaceX.

The investors who got in at 2021 will have life-changing returns, if not career defining, Gronda said. So even if you missed SpaceX in the 2010s, if you saw it somehow and got in before 2021, you're likely 20Xing your return. Founded in two thousand two, SpaceX hit unicorn status after eight years, a fast pace at the time, especially for an unproven company doing literal rocket science, Grenda noted.

Google first invested in SpaceX in twenty fifteen, joining Fidelity Investments in a$1 billion funding round that valued the company at$10 billion and gave the pair a ten percent stake. Fidelity's current stake is unknown, except for a handful of positions publicly disclosed by individual funds, end quote. Nothing more for you today, talk to you tomorrow.