Anthropic's Mythos-class model unveiled yesterday exceeds the capabilities of all its prior releases.

The big picture: Anthropic went from deeming this model so disruptive that no existing safeguards were sufficient and only a handful of organizations could access it, to releasing a public version less than three months later.

Zoom in: Don't get too excited.

• The model will be available through subscription plans for 14 days, at which point users can only access it by paying for usage credits on top of their subscriptions.

What they're saying: "The subsidy era is over" Sam Ragsdale, CEO of Merit Systems, an agentic commerce platform, wrote on X with a photo of the pricing breakdown.

• "Factor the June 22 cliff into your runway math now if you're building on it," Steffen Dybvik, an AI developer, posted.

Follow the money: Anthropic's Fable 5 — its first Mythos-class model for general use — is double the price of the AI lab's Opus models, making it the most expensive release yet.

• Anthropic stresses that expense is relative: Fable 5 also promises more intelligence and better performance, giving customers a "lower overall cost per task," Dianne Penn, Anthropic's head of product management, research and labs, told Axios.
• Anthropic says it will aim to restore Fable 5 access to standard subscription plans, "when sufficient capacity allows us to do so ... we intend to do this as quickly as we can."
• The company said that it will communicate changes ahead of time for users, but it did not provide additional clarity.

The intrigue: The release comes after the AI lab filed paperwork in preparation for an IPO.

• "These LLM companies are going to go public and they're going to raise prices because they have to," May Habib, CEO of Writer, told Axios in a March interview.
• The pressure is on to show revenue growth so strong investors want to own shares of your AI lab, not your competitor.
• That will likely require higher prices.

Frontier AI labs are converging on a new strategy for controlling their most cyber-capable models while still commercializing them: selective access.

Why it matters: OpenAI's trusted-access program and a pending program from Anthropic are creating a new power center in cybersecurity where AI companies help decide which defenders can use the most advanced cyber capabilities.

• For decades, competitive advantage in cybersecurity largely came from talent, data and infrastructure. Now, it also comes from access to models.

Driving the news: Anthropic's Fable 5 includes protections that block some high-risk cybersecurity and biology requests and instead route users who ask about those issues to Claude Opus 4.8.

• At the same time, Anthropic is offering users of its restricted Mythos Preview program an upgrade to its new Mythos 5 model.
• Anthropic's Penn tells Axios the company is being deliberately conservative at launch, meaning some legitimate security work may also get routed away from Fable 5.

The intrigue: Anthropic is also working on a formal trusted-access program that would determine who gets access to Mythos 5 and future less restricted models.

• The company has not provided a timeline for launching the program.
• Behind the scenes, organizations have spent the last two months lobbying Anthropic for access to Mythos Preview.
• Last week, the company expanded access to more than 200 companies and governments.

The big picture: OpenAI is already using a similar two-tier system.

• The company has been vetting security researchers and organizations to decide who gets access to models that could help accelerate their cyber defenses.
• The company rolled out an alternate version of its GPT-5.5 model with fewer guardrails to let those cyber defenders hunt for bugs, study malware and reverse engineer attacks.

Between the lines: It's now up to the AI labs to decide who gets access to the cybersecurity industry's most cutting-edge capabilities.

• Security vendors, researchers and critical infrastructure operators eager to get frontier AI into their products and workflows have been scrambling for access.

Reality check: Selective access allows Anthropic and OpenAI to ensure scary hacking capabilities are only in the hands of the good guys — while also finding a way to monetize their increasingly powerful models as they consider entering the public markets.

Investors have already handed the AI hyperscalers more than twice as much money in 2026 as through all of last year.

Why it matters: The astonishing scale is raising concerns over an AI bubble bursting, as well as worries over whether these investments will actually pay off in the end.

By the numbers: Just five companies — Alphabet, Amazon, Meta, Microsoft and Oracle — have raised $255.34 billion through both equity (creating new shares of stock) and debt (issuing bonds), according to data from S&P Global Market Intelligence.

What to watch: The five companies have said that by year-end they'll have spent three-quarters of a trillion dollars on AI data centers, per Barron's.

The intrigue: Four of the five hyperscalers are Big Tech companies that have long been profitable, throwing off lots of cash.

• And then there's Oracle. The nearly half-century-old software company has been on a borrowing binge to fund its AI ambitions, facing skepticism from Wall Street.
• The company reports earnings after the close today.

What's next: There's a lot more money on the way.

• SpaceX, whose ambitions for AI data centers include orbital ones, is expected to raise at least $85 billion through its IPO.
• Two more AI giants are planning stock debuts this year.

• The iPhone lowered birth rates across the U.S., according to a new paper. (Axios)
• Google is deepening its financial ties to Anthropic by backing lease payments tied to five data centers in a roughly $35 billion financing arrangement. (Bloomberg)
• A Mississippi federal judge sanctioned attorneys on both sides of a case after finding that they were using erroneous AI-generated citations to argue against each other. (404 Media)

Osiel Mendoza, who has been living with ALS since 2017, hasn't spoken directly in years, but on Monday delivered the traditional call of "Play ball" at the San Francisco Giants game using AI technology from ElevenLabs.

• ElevenLabs, which offers free voice restoration services for people with ALS, recreated Mendoza's voice using his speeches, voicemails and other recordings he made.

Watch the video.