Anthropic releases Claude Fable and Mythos 5.2 after slowdown call

Anthropic has introduced Claude Fable 5.2 and Claude Mythos 5.2. They are the company’s newest and most advanced AI models.

The release matters because it is Anthropic’s first launch of what it calls frontier models since chief executive Dario Amodei urged a slower pace of AI development in September. Frontier models sit at the leading edge of what AI systems can do.

The company is sending two messages at once. It wants to move carefully. It also wants to keep improving its products while rivals compete for the same business customers and software developers.

Think of it like a car maker that says safety testing must come first, then brings out a faster new model. Anthropic is arguing that caution does not mean stopping work.

Anthropic says the 5.2 models produce answers 30% faster than the 5.1 generation and cost 20% less. Faster results can matter when a business runs many AI tasks during the workday.

The listed price is $8 per million input tokens and $40 per million output tokens. Tokens are small pieces of text that an AI model reads or writes. Input is what a user gives the model; output is the response it creates.

Fable 5.2 is generally available. Mythos 5.2 has a narrower rollout. It is offered to approved organizations through Anthropic’s Life Sciences and Cyber Verification Programs.

That split shows how the company is separating broad commercial use from tools it sees as carrying higher risks.

Fable 5.2 is aimed at power users and demanding jobs rather than everyday requests. Anthropic says it is designed for complex, open-ended work in engineering, mathematics and robotics.

This is a different role from Claude Opus 5.5, which Anthropic released in September and positions for daily tasks. A company might use the everyday model for routine writing or summaries, then turn to Fable for a harder technical problem.

Anthropic expects some of the new capabilities to reach the Opus line over time. For now, it says the performance gap makes the newer model better suited to these frontier uses.

The goal is clear: make the product useful enough that developers and companies choose it for work where a weaker tool could waste time or produce a poor answer.

Anthropic says both models went through rigorous safety and alignment testing. Alignment means checking whether an AI system behaves in ways intended by its maker.

The company also developed new evaluations and tools to monitor the models after launch. This matters most for Mythos, whose advanced cybersecurity abilities led Anthropic to limit access when the earlier version arrived in April.

Fable first appeared two months later as a public version of Mythos with added safeguards. The 5.2 release keeps that basic structure: wider access for Fable and controlled access for Mythos.

Anthropic’s position is that it called for pacing, not a pause. The new release puts that claim into practice, but it also raises the pressure to show that the safeguards work as the tools reach more users.

The launch comes as AI companies fight to become a regular supplier to large organizations. Winning that business is not only about headline-grabbing model scores. It is also about speed, cost, reliability and clear rules for using powerful tools.

Anthropic reported $33.24 billion in revenue for the first nine months ended September 30, up roughly 1,200% from a year earlier. That growth gives the company a strong reason to keep bringing new products to market.

In broader trading, the technology-heavy QQQ exchange-traded fundETFA fund traded on the exchange like a share. It usually tracks an index, so one purchase buys a whole basket of companies at low cost.Full explanation in the glossary → was up 3.29% at the time of writing. Nvidia shares, a major AI-chip bellwether, were down 5.38%. Those moves do not measure demand for Anthropic’s new models, but they show how closely markets are watching the AI industry.

For Anthropic, the next test is practical. Businesses and developers will judge whether Fable 5.2 delivers useful gains, while the company tries to keep its safety promises credible.