Anthropic Goes Public With Its Most Powerful AI

Anthropic has launched Claude Fable 5, a Mythos-class AI model now widely available to enterprise customers and paid subscribers — two months after its highly restricted debut captivated Wall Street and government officials alike.

The original Mythos model, unveiled in April, was kept under tight control due to its advanced ability to identify software security vulnerabilities. Access was limited to a select group of organizations under Anthropic's cybersecurity initiative, Project Glasswing. Now, with new safeguards in place blocking high-risk responses across areas like cybersecurity and biology, Anthropic believes broader deployment is responsible and ready.

Claude Fable 5 delivers exceptional performance across software engineering and knowledge work, scoring over 10% higher than Claude Opus 4.8 on key benchmarks. When a dangerous query is detected, the model automatically defaults to Opus 4.8 for a safe response — a deliberate, layered safety architecture.

Alongside Fable 5, Anthropic released Claude Mythos 5 - the same underlying model with select safeguards lifted — exclusively for trusted partners.

Pricing reflects the model's premium capability: $10 per million input tokens and $50 per million output tokens, double that of Opus 4.8. Early users report improved ROI per task, validating the higher price point.

The launch comes days after Anthropic confidentially filed its IPO prospectus with the SEC. With a reported $47 billion revenue run rate and a $965 billion valuation — surpassing OpenAI's $852 billion — Anthropic is positioning Claude Fable 5 as a landmark product ahead of what could be a historic public market debut.