Artificial intelligence company Anthropic has released a new advanced AI model, Claude Fable 5, as part of its “Mythos-class” system rollout, describing it as its most powerful model yet for general users.
The company also introduced Claude Mythos 5, a more capable version of the system that will remain restricted to trusted partners due to safety and cybersecurity concerns.
Claude Fable 5, which is now available to the public, is built on the same core architecture as Mythos 5 but includes stricter safety guardrails designed to limit misuse in sensitive areas such as cybersecurity, biology and chemistry.
According to Anthropic, when users attempt restricted queries, the system can automatically redirect requests to a less capable but safer model, Claude Opus 4.8.
The company says the model is designed for high-level software engineering, research, and complex analytical tasks, and is capable of handling longer and more autonomous workflows compared to previous versions.
Claude Fable 5 is priced at $10 per million input tokens and $50 per million output tokens, making it one of the company’s most expensive public AI offerings.
The release follows earlier limited access testing under Anthropic’s “Project Glasswing” initiative, which included government and selected private-sector partners. Reports suggest the Mythos system was previously only available to a small number of organisations working in cybersecurity and research environments.
Anthropic says the dual-release approach is intended to balance broad access with controlled deployment of more powerful AI systems.
The launch comes amid growing global debate over how frontier AI models should be regulated, particularly as they become increasingly capable of identifying software vulnerabilities and performing complex reasoning tasks.
No timeline has been announced for wider public access to Claude Mythos 5.