Anthropic's Claude Fable 5 is here: The Mythos-class AI model anyone can now use and what makes it different

Anthropic has launched Claude Fable 5, the first publicly available version of its Mythos-class AI model. The launch of this AI model comes roughly two months after Mythos was first introduced as a preview in April, when access was limited to a small group of partners, under Project Glasswing, due to cybersecurity concerns. According to the company, Fable 5’s capabilities exceed those of any model Anthropic has previously made available to general public. The model now accessible to its enterprise clients and paid subscribers.

“For us, it’s really around what we call ‘race to the top,’ being able to provide this technology in a valuable fashion, and at the same time providing the right safety guardrails so that it can do asymmetrically more benefits than harm,” CNBC quoted Dianne Penn, Anthropic’s head of product management for research.

What Anthropic Claude Fable 5 does

Anthropic noted Claude Fable 5’s state-of-the-art performance across software engineering, knowledge work, vision and scientific research. To address misuse risks, the model ships with safeguards that redirect certain high-risk queries, particularly in cybersecurity, to Claude Opus 4.8 instead. Anthropic says it has figured out how to safely hand that same ‘god-like’ computing power to the general public.

What makes Claude Fable 5 different from Mythos

The primary difference between the hidden Mythos model and the newly released Claude Fable 5 comes down to a digital security limitation. In Claude Fable 5, Anthropic has built an advanced architecture designed to block responses in high-risk areas like biochemical threats and cybersecurity.

Fable 5 uses real-time safety classifiers. If a user tries to exploit the model by asking for instructions on how to manufacture a toxin like ricin, Fable 5 will immediately freeze, block the response and quietly hand the prompt over to a less powerful, standard model (Claude Opus 4.8) to deliver a safe answer.

Anthropic also unveils Claude Mythos 5

Further, for specialised security professionals who need the guardrails lifted, Anthropic also launched Claude Mythos 5 – the exact same model as Fable 5 but with the defensive boundaries removed for certified enterprise tasks. Citing internal benchmarks, Anthropic says that Claude Fable 5 represents a massive leap forward in logic, software engineering, and complex knowledge work, scoring more than 10% higher than the company's previous flagship model released just last month.

However, that elite intelligence comes with a steep price tag. Fable 5 is priced at $10 per million input tokens and $50 per million output tokens, which essentially makes it twice as expensive to run as Claude Opus 4.8. Anthropic claims the model’s accuracy actually saves companies money.

“You just get a higher ROI (Return on Investment) by having more intelligent models,” Penn explained, noting that early corporate testers reported a much higher success rate per task, reducing wasted computing costs.

The release of Claude Fable 5 comes days after Anthropic confidentially filed its prospectus with the SEC for a historic Initial Public Offering (IPO) later this year. Anthropic revealed its revenue run rate has skyrocketed to $47 billion, up from $10 billion last year. A recent funding round valued the company at $965 billion, officially pushing it past its chief rival, OpenAI, which was valued at $852 billion in March.