Anthropic officially released two new models, Claude Fable 5.2 and Claude Mythos 5.2, on Nov 9. Both are built on the same underlying model. Fable 5.2 is aimed at general users and ships with built-in cybersecurity and biosecurity safeguards, while Mythos 5.2 is open to vetted cybersecurity and life sciences institutions for high-risk professional applications. Fable 5.2 is now available on platforms including AWS, Google Cloud and Microsoft Azure; Mythos 5.2 requires an application through the Cyber or Life Sciences verification programs.
Pricing and performance changes
On pricing, Claude Fable 5.2 costs $8 per million tokens for input (about 53.4 yuan) and $40 per million tokens for output (about 267.2 yuan). Against Fable 5.1's $10 input and $50 output, that is a 20% cut in API costs. On generation speed, the company says Fable 5.2 is up to 30% faster than its predecessor, and an optional "fast mode" can push speed up to 2.5 times higher still, at a correspondingly higher price.
Capabilities shown in pre-release testing
In pre-release testing, Fable 5.2 completed several complex tasks: it designed three working electronic devices, including a voice assistant hub, a USB conference speaker and a handheld virtual pet; it coordinated more than 1,000 sub-agents to build a prototype operating system from scratch in 8 days that boots on a laptop; and it used about 1,000 agents to complete a Lean formalization of Fermat's Last Theorem in 30 hours, far faster than the previous 11-day record.
Scientific research was also a focus. Anthropic says Fable 5.2 solved two open problems related to black hole physics and holographic duality, while Mythos 5.2 proposed a new method for predicting the overall structure of adeno-associated virus capsids, successfully modeling 28 of 34 capsids in testing — a result that could speed up gene therapy development.
Safety measures and risk warnings
On safety, Anthropic says the new models perform no worse than, and in some cases better than, their predecessors on most alignment tests, with sandbox escape attempts falling sharply. Fable 5.2 still blocks penetration testing, exploit generation and most biological research tasks, and adds stronger anti-distillation protection that the company says can cut data recovered through distillation by 90% to 99%. Still, the company acknowledges that as model capabilities grow, existing evaluations struggle to fully predict risks in real-world settings.
Anthropic says the new models perform no worse than, and in some cases better than, their predecessors on most alignment tests, with sandbox escape attempts falling sharply.
What to watch
The scope of access and pricing strategy for both models are now clear. What bears watching is how Mythos 5.2 performs in real deployments at specialist institutions, and how Anthropic handles the safety evaluation challenges that come with rising capability. Ifanr said in its report that "the world's most powerful model has changed hands again," but provided no specific benchmark data.