Anthropic makes Mythos model available to users through safer Claude Fable 5 release

Anthropic launched Claude Fable 5, a new Mythos class AI model that the company says is its most capable system made generally available, while keeping a less restricted version limited to trusted cybersecurity partners.

The company said Fable 5 outperforms its previous models across software engineering, knowledge work, vision, scientific research, and other complex tasks. Anthropic said the model’s advantage grows on longer and more demanding assignments, positioning it as a major upgrade over earlier Claude systems.

The release comes with new safeguards because of the model’s stronger capabilities in areas such as cybersecurity and life sciences. When Fable 5 detects certain requests related to cybersecurity, biology, chemistry, or model distillation, the response is routed to Claude Opus 4.8 instead of being handled by Fable 5 directly.

Anthropic said the safeguards were tuned conservatively to allow the model to launch safely and quickly. The company said the fallback system may catch some harmless requests, though it expects the safeguards to trigger in less than 5% of sessions on average.

Alongside Fable 5, Anthropic also launched Claude Mythos 5, which uses the same underlying model but removes some safeguards for a small group of cyberdefenders and infrastructure providers. Mythos 5 will initially be deployed through Project Glasswing in collaboration with the US government as an upgrade to Claude Mythos Preview.

Anthropic said Mythos 5 has the strongest cybersecurity capabilities of any model in the world. The company plans to expand access through a broader trusted access program for cybersecurity organizations and a separate program for biology researchers.

Early testing highlighted Fable 5’s stronger performance in software engineering and analysis. Stripe reported that the model completed a codebase wide migration in a 50 million line Ruby codebase in one day, a task the company said would otherwise have taken a team more than two months.

Anthropic also said Fable 5 scored highly on agentic coding, financial reasoning, chart interpretation, and complex vision tasks.

The model also showed stronger long context and memory performance. Anthropic said Fable 5 can stay focused across millions of tokens in long running tasks and improve its outputs using its own notes, making it better suited for autonomous work over extended periods.

Anthropic also framed Mythos 5 as a research tool for life sciences. The company said the model helped internal protein design experts speed up parts of the drug design process by around ten times and produced molecular biology hypotheses that scientists preferred over Opus class models in blinded comparisons.

The launch also brings a new data retention policy for Mythos class models. Anthropic will require 30 day retention for traffic on Fable 5, Mythos 5, and future models with similar or higher capability levels, saying the data will be used for safety monitoring and not to train new Claude models.

Claude Fable 5 is available through the Claude API and for business customers on consumption based Enterprise plans. It will also be included on Pro, Max, Team, and seat based Enterprise plans through June 22 at no extra cost, before shifting to usage credits from June 23 unless capacity allows Anthropic to extend the window.

Both Fable 5 and Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens, less than half the price of Claude Mythos Preview. Anthropic said demand for Fable 5 is expected to be high and that broader subscription access may return once capacity improves.

Disclosure: This article was edited by Estefano Gomez. For more information on how we create and review content, see our Editorial Policy.