Anthropic has made its Mythos‑class model available to the public with Claude Fable 5.
The company released the new model on Tuesday for enterprise clients and paid subscribers. It uses the same architecture as the restricted Mythos system, but includes filters that block answers related to cybersecurity, biology, chemistry and a few other sensitive areas. Access is still limited to paying users; there is no free tier.
This matters because it is the first time a Mythos‑level model can be run outside Anthropic’s private labs, potentially raising the baseline for commercial AI performance. At the same time, the added safeguards show Anthropic’s effort to curb misuse while courting regulators.
The timing is curious: the launch arrives just days before the company’s rumored record‑size IPO, giving investors a concrete product to evaluate.