Anthropic is releasing Claude Fable 5, described across outlets as the first “Mythos-class” Claude model available to the general public. Multiple reports say the company had previously restricted the full Claude Mythos 5 due to safety concerns, and is making a more limited variant accessible instead. Sources agree that Fable 5 is built on the same underlying Mythos architecture, but includes safeguards intended to reduce misuse in high-risk domains. TechCrunch, ZDNet, Help Net Security, Forbes, and others report that the model blocks or limits responses related to cybersecurity and scientific areas such as biology, with some coverage also mentioning chemistry. ZDNet and The Next Web also note additional product measures, including fallback behavior (such as switching to other responses/models) and availability through enterprise customers and paid subscribers. Several outlets frame the release as bringing advanced coding and research-oriented capability to non-restricted users while maintaining guardrails, which Anthropic says are necessary because Mythos-class systems could otherwise provide guidance that could be used for serious harm.