Anthropic has spent years arguing that powerful AI and responsible deployment can coexist. Its handling of two models launched in 2026 is the clearest test of that argument yet. Claude Mythos and Claude Fable 5 represent the frontier of what AI can currently do, and Anthropic has made a deliberate, uncommon choice about who gets access to them.

The reasoning is not about commercial advantage. It is about risk.

What Claude Mythos and Fable 5 Actually Are

Claude Fable 5 is a Mythos-class model that Anthropic has made safe for general use. Its capabilities exceed those of any model Anthropic has previously made available. It is state-of-the-art on nearly all tested benchmarks, showing exceptional performance in software engineering, knowledge work, vision, scientific research, and many other areas.

Key distinctions between the two models:

  • Claude Fable 5 is generally available on the Claude API, Claude Platform on AWS, Amazon Bedrock, Vertex AI, and Microsoft Foundry. Claude Mythos 5 is not generally available and is offered only in limited availability to approved customers in Project Glasswing.
  • Claude Mythos 5 is essentially Fable 5 without many of the safeguards and is accessible only to a small group of cyberdefenders and infrastructure providers.
  • Claude Fable 5 includes safety classifiers that can decline requests. Claude Mythos 5 does not include these classifiers.

Anthropic's early data shows that more than 95% of Fable 5 sessions involve no fallback at all, meaning for most users, Fable 5's performance is effectively the same as that of Mythos 5.

The Cybersecurity Threat That Changed Anthropic's Release Strategy

Claude Mythos Preview has already found thousands of high-severity vulnerabilities, including some in every major operating system and web browser. Given the rate of AI progress, it will not be long before such capabilities proliferate, potentially beyond actors committed to deploying them safely.

That finding reshaped how Anthropic approached the release entirely.

  • Project Glasswing bundles restricted access to Claude Mythos Preview with up to 100 million dollars in usage credits and 4 million dollars in donations to open-source security groups.
  • Launch partners include Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks.
  • Anthropic has since broadened access, adding approximately 150 organizations across 15 countries as part of Project Glasswing's expansion.

For public users, Fable 5 includes guardrails that redirect high-risk queries, such as requests involving dangerous substances or cybersecurity exploits, to Claude Opus 4.8 to deliver a safer response instead.