This website uses cookies

Read our Privacy policy and Terms of use for more information.

Opus 4.7 gets 2.5x faster

Welcome back! Anthropic announced that Opus 4.7 is going to be the default for fast mode, which costs higher than the standard rate per token, so leaving the toggle on will run up your extra-usage bill fast. We have another Musk vs. Altman update, with new testimony that could change the trajectory of the case. Google is launching a new laptop this fall with a new AI gimmick and Meta is losing the support of its employees over the mouse-tracking software it installed last month.

In today's Generative AI Newsletter:

  • Speed economy: Why is Anthropic putting a premium on the same Opus running faster?

  • OpenAI trial: What changes when Altman tells a jury Musk wanted 90 percent?

  • Google's laptop bet: What did Google build into the cursor itself?

  • Meta revolt: Who owns the data Meta is using to train its next model?

Anthropic announced Opus 4.7 is going to be the default for Fast Mode on its API and in Claude Code, running the model at 2.5x its standard speed. The feature is in research preview and switches on with /fast inside Claude Code.

The details:

  • Pricing: $30 per million input tokens and $150 per million output tokens, flat across the 1M context window.

  • Quality: Fast Mode uses the same Opus weights and capabilities with a different API configuration that prioritizes speed over cost efficiency.

  • Partners: Fast Mode for Opus 4.7 is also live in Cursor, Emergent, Factory, v0, Warp and Windsurf.

Nothing new here, Anthropic is treating speed as a product line. It’s the same model… but faster and with a higher price per token. The Thursday default switch means anyone running /fast is about to see their Opus get smarter and WAY more expensive in the same instant.

Special highlight from our network

Apple's linking iPhones to Starlink. Your phone is now your money-maker anywhere you go.

Financial stress is universal: most people can't handle a $1,000 surprise bill, and billions earn under $12,000 yearly with no backup plan.

Mode Mobile solved it. Their EarnPhone turns everyday phone habits into income: stream music, play games, charge your device and get paid.

With 490M+ users and a combined $1B in earnings and savings, the proof is in the numbers.

Invest now.

Disclaimer

Please read the offering circular and related risks at invest.modemobile.com. This is a paid advertisement for Mode Mobile’s Regulation A+ Offering. Mode Mobile recently received their ticker reservation with Nasdaq ($MODE), indicating an intent to IPO in the next 24 months. An intent to IPO is no guarantee that an actual IPO will occur.

The Deloitte rankings are based on submitted applications and public company database research, with winners selected based on their fiscal-year revenue growth percentage over a three-year period.

We started by helping people learn about AI. Two years on, that grew into a 14M+ community, 300+ paying customers and 2.4x revenue in 2024-2025.

Now we're entering the next stage, moving from sharing AI to teaching it so people can apply it in their work. We do that through our ecosystem of products, and they're already gaining traction. AI Academy launched with 30K students and $12K in month one. ToneUp and the Hackathon app come next.

To fund this stage, we opened a crowdfunding round. Join from $1,000. Up to 25% bonus shares before July 1, 2026.

In making an investment decision, investors must rely on their own examination of the issuer and the terms of the offering, including the merits and risks involved. Genai Works, Inc. has filed a Form C with the Securities and Exchange Commission in connection with its offering, a copy of which may be obtained here.

Sam Altman took the witness stand on Tuesday and pushed back against Elon Musk's allegation that he betrayed OpenAI's nonprofit mission. Altman testified that Musk himself had asked for a 90 percent equity stake in the early days of the company and pressed for majority control throughout.

The details:

  • The number: "An early number that Mr Musk threw out was that he should have 90 percent of the equity to start. It then softened, but it always was a majority."

  • The succession question: Asked what would happen to OpenAI after his death, Musk said he might pass it to his children. Altman called that a "hair-raising moment."

  • Damages: Musk is seeking $150 billion from OpenAI and Microsoft, payable to an OpenAI nonprofit. He also wants Altman and Brockman removed and the company reverted to nonprofit status.

  • Cross-examination: Musk's lawyer Steven Molo cited a former board member's claim of a "toxic culture of lying" and pressed Altman on whether he had misled people in business. Altman denied it.

Musk's case has always rested on the framing that he was duped into funding a charity that turned for-profit and Altman's 90 percent line cuts through that framing in front of a jury. 

OpenAI is preparing for a possible IPO that could value it at a trillion dollars, and the trial now sits on the path between today and that public offering.

Google announced Googlebook today, a new laptop line running a hybrid of Android and ChromeOS with Gemini built into the operating system. 

The marketing centerpiece is Magic Pointer, a cursor that summons Gemini when you wiggle it. 

Point at a date inside an email and Gemini offers to schedule the meeting. 

Hover a couch image while a photo of your living room is open and Gemini composites the two. 

The deeper move is what Google is not saying out loud. Googlebook runs on the Android tech stack with ChromeOS folded inside, and that is Google admitting ChromeOS as a separate operating system is closing out. 

The Chromebook era of cheap web boxes for schools is over, and the same OEM roster that built those Chromebooks for a decade is now being asked to ship premium hardware that competes with Microsoft's Copilot+ PCs and Apple Intelligence Macs.

Magic Pointer is the feature that justifies the rebrand. But it makes you wonder. Couldn't this have been a software update that works across any laptop? Let's keep an eye on what reviewers like Mrwhosetheboss and Marques Brownlee say when it drops this fall.

Meta employees distributed flyers across multiple US offices on Tuesday in protest of the mouse-tracking and keystroke-logging software the company installed on employee machines last month. 

The flyers appeared in meeting rooms, on vending machines and on top of toilet paper dispensers, pointing colleagues to an online petition. The protest lands about a week before Meta begins planned layoffs of 10 percent of its workforce.

Meta is framing the data collection as AI training material, and that framing is what makes the story matter. 

The company is harvesting employee work, cursor movements, keystrokes and behavioral patterns, to train models that will eventually do the jobs of the people generating the data. Then it is laying off 10 percent of them.

The AFL-CIO poll in Light Bytes below describes how American workers feel about this scenario, and Meta has just handed them a textbook example. Watch the phrase "AI training data." It is the corporate language that turns workplace surveillance into a model alignment exercise.

Gumloop is a no-code platform for building AI agents that run inside the tools your team already uses, with a drag-and-drop canvas and prebuilt templates for sales, support, data analysis and meeting prep. It lets you call an agent from a Slack channel by tagging @Gumloop the way you would a colleague.

Try this yourself:

  • Sign up at gumloop.com and click on Get Started.

  • Use one of the existing templates.

  • Connect one service you use every day (Slack, Notion, Salesforce or Gmail) and run the agent against real data.

  • Claude for Legal teams: Anthropic rolled out Claude for Legal with practice-area plugins for commercial, employment, privacy and corporate work, plus MCP connectors into DocuSign, iManage, LexisNexis and Thomson Reuters, with Freshfields already standardized on it.

  • OpenAI and Microsoft cap revenue-share at $38B: The two companies have agreed to cap revenue-sharing payments at $38 billion, freeing OpenAI to deepen ties with Amazon and Google ahead of a possible IPO.

  • AI guardrails on the Trump-Xi summit agenda: Washington and Beijing both name AI-enabled bioweapons and infrastructure attacks as shared concerns ahead of this week's Beijing summit.

  • 95% of US workers want a human in the loop: A new AFL-CIO poll shows 95 percent of workers favor job and privacy protections from artificial intelligence.

  • Demi Moore tells Cannes the film industry cannot win against AI: The actor and jury member said fighting AI is a battle the industry will lose.

Reply

Avatar

or to participate

Keep Reading