Anthropic releases “Claude Fable 5,” its first Mythos-class AI model made available to the general public. Multiple outlets report that Anthropic previously described the full “Claude Mythos 5” as too risky to release publicly, and now offers a more restricted alternative that is designed for everyday use. According to the sources, Fable 5 uses the same underlying Mythos-class architecture or capabilities as the restricted Mythos system, but includes safety measures to limit misuse.
The reports agree that Claude Fable 5 adds guardrails that block or restrict responses in high-risk domains, particularly cybersecurity-related content and areas such as biology (and in some accounts, chemistry as well). Some coverage also notes that the model can use fallback behavior or fallback models when requests fall into restricted categories. Availability is described as including enterprise customers and paid subscribers rather than a fully open release. Overall, outlets characterize Fable 5 as bringing Mythos-level capability to public users while keeping stronger constraints than the unreleased or restricted Mythos model.
