MAIN ARTICLE
Two products built on the same foundation
The distinctive feature of the release is that Fable 5.2 and Mythos 5.2 are not separate underlying intelligence systems. Anthropic says they share a foundational model while applying different safety controls and eligibility requirements. Fable is intended for a wide audience tackling programming, complex reasoning, engineering and multi-agent coordination. Mythos retains capabilities related to cybersecurity and biological research, but access is limited to organizations approved through specialized verification programs.
That arrangement reflects a growing dilemma for the industry. Systems capable of locating software vulnerabilities or assisting sophisticated scientific research have valuable legitimate uses, but some capabilities also carry misuse risks. Screening access cannot eliminate every danger, and it raises questions about who sets the rules. Still, the distinction between the two products signals that releasing a frontier model is becoming an exercise in managing real-world consequences, not merely publishing a more capable chatbot.
Lower prices and faster responses need context
Anthropic reports output generation up to 30% faster than Fable 5.1 and Mythos 5.1. It set standard prices at $8 for one million input tokens and $40 for one million output tokens, down from the prior $10 and $50.

















