June 9, 2026

Anthropic Claude Fable 5 and Mythos 5 released

AnthropicAn AI safety company — creators of the Claude family of language models. released Claude Fable 5 and Mythos 5 on June 9, 2026 — its fifth-generation frontier models, split into a generally available safe build and a restricted cyberdefense variant.

What it was for

AnthropicFable 5 and Mythos 5 share the same underlying model. Fable 5 is AnthropicAn AI safety company — creators of the Claude family of language models.'s top generally available tier: safety classifiers route sensitive cybersecurity and biology queries to Claude Opus 4.8 instead. Mythos 5 lifts those safeguards for vetted partners in Project Glasswing — AnthropicAn AI safety company — creators of the Claude family of language models.'s answer to releasing a model strong enough to help defenders without open-sourcing misuse risk. Both ship at $10/$50 per million tokens.

Companies

  • Anthropic

Why it's here

AnthropicAn AI safety company — creators of the Claude family of language models. split one frontier model into public Fable and restricted Mythos to balance capability and safety.

Why it mattered

It defined how labs ship models with dangerous dual-use skills without a single open release.

What it solved

Frontier cyber and biology capability needed deployment paths for defenders, not just blanket refusals.

Media

  • Anthropic
    ImageAnthropic

    Anthropic, Public domain, via Wikimedia Commons

Related