Anthropic Unveils New Claude AI Models Fable 5 and Mythos 5

Anthropic has announced its latest generation of AI models, Claude Fable 5 and Claude Mythos 5. Fable 5 is designed for general knowledge work and coding, while Mythos 5 demonstrates advanced capabilities in specialized fields like cybersecurity and biology.
Anthropic Unveils New Claude AI Models Fable 5 and Mythos 5

Anthropic Unveils New Claude AI Models Fable 5 and Mythos 5 Anthropic’s latest AI announcement has quickly evolved from a showcase of cutting‑edge capability into a live test of access, safety, and reliability for its most advanced models.

On June 9, 2026, the company detailed Claude Mythos 5, describing it as its “most capable model for cybersecurity and biology research,” emphasizing major gains in cybersecurity, biology, and healthcare benchmarks. Because these strengths are highly dual‑use, Anthropic restricted Mythos 5 to “a small group of vetted partners,” with plans to expand availability over time.

In parallel, Anthropic introduced Claude Fable 5, billed as the “next generation of intelligence for the hardest knowledge work and coding problems.” Fable 5 shares Mythos‑level capabilities but is tuned for days‑long, complex tasks, autonomous planning, and self‑correction across coding and enterprise workflows. It is positioned as the general‑purpose counterpart to Mythos, with the same underlying model but different safety and routing behavior.

The company framed the two‑track design as a safety measure: in sensitive areas such as cybersecurity and biology, Fable 5 automatically routes queries to the earlier Claude Opus 4.8 model, while Mythos 5 remains tightly controlled through “trusted access programs.”

However, by June 12, 2026, both launches were overshadowed by availability setbacks. Anthropic issued updates noting that “Claude Mythos 5 is currently unavailable” and that access for both Mythos 5 and Claude Fable 5 was disrupted for Glasswing partners, with apologies and assurances it was working to restore service. A similar note confirmed that “Claude Fable 5 is currently unavailable,” despite published pricing of $10 per million input tokens and $50 per million output tokens.

The result is a layered rollout: powerful, benchmark‑leading models with tightly partitioned safety regimes, but also early operational instability that leaves prospective users weighing promise against immediate access and reliability concerns.

Continue reading https://foxvector.com/stories/019f0f43-8763-0a05-71de-13a088da2d39

Write a comment