Anthropic has released Claude Fable 5, a consumer-facing version of the Mythos-class AI technology the company spent months refusing to make public over safety concerns. The model is now available on Claude Pro, Max, Team, and Enterprise subscription plans at no additional cost, though access conditions are set to change later this month.
Until now, Anthropic had kept Mythos-class AI internal, citing concerns about its potential in cybersecurity and biological research. The company had repeatedly described the technology as requiring extraordinary safeguards before any public release.
What Claude Fable 5 can do
According to Anthropic, Fable 5 is built for extended, autonomous work rather than single-turn conversations. The company says it can handle tasks that unfold over hours or days, including software engineering projects, in-depth research, and complex AI agent workflows. Rather than waiting for each new prompt, the model can work through multi-step problems with greater independence.
The improvements appear aimed primarily at power users, researchers, and developers, rather than casual users asking questions or drafting emails.
How Anthropic is managing the risk
Anthropic acknowledges that “releasing a model this capable comes with risks.” Its main safeguard is a system of AI classifiers that run alongside Fable 5 and monitor incoming requests for signs of misuse. When those classifiers detect a request related to advanced cybersecurity activity, the system routes it to Claude Opus 4.8 instead, keeping the most sensitive capabilities away from the more powerful model.
The company claims Mythos has “the strongest cybersecurity capabilities of any model in the world.” The same tools that help security researchers identify software vulnerabilities could, in the wrong hands, be used to exploit them.
Access and pricing
From today, Fable 5 is available on Claude Pro, Max, Team, and seat-based Enterprise plans at no added charge. From June 23, however, access shifts to a usage-credit system. Subscribers who want to continue using Fable 5 after that date may need to purchase additional credits depending on how much they use the model.
Anthropic says the change is temporary. “After this point, when sufficient capacity allows us to do so, we aim to restore Fable 5 as a standard part of subscription plans,” the company said. “We intend to do this as quickly as we can.”
Early users have flagged the model’s token usage. One Reddit user on the Max 20x plan reported usage climbing at roughly 2% per minute during heavier sessions.
Not all reactions have been positive. One user described the restriction of cybersecurity capabilities as “a preview of AI inequality,” arguing the public receives a safer but less capable version while trusted institutions get access to the full model.
The release puts Anthropic in an unusual position, having spent months describing Mythos as too dangerous to release publicly and now placing it, with guardrails, in the hands of ordinary subscribers.