It’s only been 11 days since Claude Opus 4.8 hit the scene, and Anthropic, the world’s most valuable AI startup, has already unveiled its latest foundation model.
On June 9 (ET), Anthropic officially rolled out its next-gen models: Claude Fable 5 and Claude Mythos 5. They’re built on the exact same underlying architecture. Think of Fable 5 as the public-ready version of the Mythos tier, while Mythos 5 holds back even more raw power and is currently locked down to a handful of vetted partners. Same day, Anthropic also dropped the pricing: both models will run you $10 per million input tokens and $50 per million output tokens.
According to Anthropic, Claude Fable 5 is hands-down the most capable general-purpose model the company has ever publicly released, topping benchmarks in software engineering, scientific research, financial analysis, and visual reasoning. But here’s the catch: Anthropic intentionally dialed back some permissions for Fable 5. If your prompt brushes up against sensitive territory like cybersecurity, biology, or chemistry, the system automatically routes it to the slightly less powerful Claude Opus 4.8 to keep misuse off the table.
The team says Fable 5 crushes nearly every major benchmark, especially when it comes to long-running, multi-step projects that demand serious heavy lifting. Software dev is where it really shines. Take Stripe, for example: early testers put Fable 5 to work on a massive 50-million-line Ruby codebase. What would normally take a dedicated team two full months? The model knocked it out in a single day.
Beyond Fable 5, Anthropic also dropped a bombshell: despite sharing the same base architecture, Mythos 5 runs with noticeably fewer safety guardrails. Access will be tightly controlled through a special initiative called Project Glasswing. We’re talking strictly vetted partners like U.S. government agencies, critical infrastructure operators, and elite cyber defense units.
Anthropic notes that Mythos 5 can now handle end-to-end protein design workflows, punching above its weight in certain drug discovery tasks—even outperforming seasoned researchers. In fact, the Anthropic research team shared that a brand-new mechanism hypothesis the model proposed for an E. coli protein was later independently validated by outside labs.
Anthropic is also betting big on security, claiming Mythos 5 boasts the “world’s strongest cybersecurity capabilities“. Going forward, the company plans to slowly open the floodgates via its “Trusted Access Program“.
Ironically, Anthropic has been loudly advocating for stricter AI safety lately. Just a few days prior, on June 4, they published a deep dive calling for a global pause on cutting-edge AI development. Co-authored by co-founder Jack Clark and head of research Marina Favaro, the paper leans hard on internal metrics to prove that “recursive self-improvement (RSI)“ is accelerating faster than anyone anticipated. That reality check forms the backbone of their push for coordinated global oversight.
Inside the house, the numbers are staggering. By May 2026, a whopping 80% of shipped code at Anthropic is fully autogenerated by Claude. Compared to the same window in 2024, engineers are pumping out eight times more functional code per day. When you slot in the latest Mythos Preview build, research productivity jumps fourfold versus working without AI. Even more impressive? The time it takes to double the complexity of tasks AI can tackle solo has halved from seven months down to just four. In brutal algorithm optimization scenarios where humans might sweat for hours to squeeze out a 4x speed boost, Mythos clears that gap in one shot, delivering a massive 52x efficiency jump.
So while they’re publicly asking rivals to hit the brakes, Anthropic is absolutely flooring it on its own release cycle. This latest drop came just 11 days after the last one. For perspective, the May 28 rollout of Claude Opus 4.8 followed its predecessor, Opus 4.7 (released April 16), by a much more relaxed 41 days.
Right around the same time, Anthropic has been quietly ramping up fundraising efforts and prepping for a major IPO.
Back on May 28, Anthropic closed a jaw-dropping $65 billion Series H round, catapulting its post-money valuation to $965 billion—officially dethroning OpenAI’s $852 billion mark. If we rank this against S&P 500 giants, Anthropic is already sitting pretty as the 13th largest company in the U.S. To put that in perspective, Nvidia leads the pack at $5.18 trillion, with Google, Apple, and Microsoft hot on its heels. Meanwhile, Anthropic has already surpassed Intel and Walmart, trailing only Berkshire Hathaway.
In their official statement, Anthropic made it clear: this war chest goes straight into strengthening AI safety and transparency research. They’ll also be massively scaling compute infrastructure to keep up with skyrocketing Claude usage, while expanding their product suite and partner ecosystem.
Roundtable investors include Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital. The deal also locks in previously pledged cloud commitments totaling $15 billion, with Amazon shelling out $5 billion alone. Adding fuel to the fire, memory chip titans Micron, Samsung Electronics, and SK hynix all signed on to the investor list.
A quick refresher: Anthropic kicked off in 2021. Unlike OpenAI, which tries to serve both consumers and businesses, Anthropic has kept its sights firmly on enterprise buyers. With Claude’s massive 200K token context window, rock-solid security, and beefy reasoning chops, it’s quickly become the go-to deployment engine for companies serious about AI.
Financials are looking incredible too. As of April 2026, Anthropic’s annualized revenue smashed past $47 billion—a healthy leap over $50% from the $30 billion mark right after their G round in February. That translates to over $5 billion in fresh monthly revenue. And because almost all of that cash flows through enterprise API calls across high-value sectors like finance, legal, and tech, customer stickiness is incredibly tight.
Word first surfaced back in March that Anthropic is eyeing a public listing as early as October, potentially pulling in over $60 billion during the debut. Reports confirm they’ve already filed a confidential draft with the SEC in early June, though share counts and pricing remain under wraps. Goldman Sachs and Morgan Stanley are lined up as lead underwriters, with JPMorgan also jumping in. Insiders hint there’s still room to bring on additional banks before the tape starts trading.
When asked about the road to going public, Anthropic co-founder and president Daniela Amodei pointed out that the sheer price tag of training next-gen models is basically forcing companies like hers to tap the public markets. “Building advanced AI models is an incredibly capital-intensive grind,” she explained. “The public markets are just tailor-made for what we’re doing.”
