Anthropic’s Claude Fable 5 is out for public use, with safeguards for high-risk requests
Days after publishing research on how advanced AI systems could amplify cyber operations in the wrong hands, Anthropic released Claude Fable 5, a Mythos-class model for general use.
“Releasing a model this capable comes with risks. Without safeguards, Fable 5’s capabilities in areas like cybersecurity could be misused to cause serious damage,” Anthropic wrote.
The company said Mythos-class models possess advanced cybersecurity and research biology capabilities that can provide information and guidance beyond what is typically available through conventional online sources. Those capabilities could assist malicious actors in carrying out harmful activities while also supporting legitimate work by cybersecurity professionals and researchers.
To address those risks, Fable 5 comes with a set of classifiers designed to detect requests involving cybersecurity, biology and chemistry, and model distillation.
When those classifiers are triggered, requests are automatically routed to Claude Opus 4.8 and users are informed. Anthropic noted that the safeguards activate in less than 5% of sessions on average.
“Over the past few months we have been improving these safeguards, and they are now robust enough for a general release,” the company noted.
Fable 5’s safeguards underwent extensive internal red-team testing and an external bug bounty program. Anthropic said more than 1,000 hours of testing did not produce a universal jailbreak.
Industry reaction
Chris Boehm, field CTO at Zero Networks, views the release as both a competitive move and an effort to make advanced AI capabilities available to a broader audience.
“Part of it is competitive, plain and simple. Being first to put a model like this in the market matters, and they want that flag in the ground,” Boehm said. “The hard part was never building the capability; it was figuring out how to get it into customers’ hands at scale without it being reckless.”
He also highlighted wider availability as one of the most significant aspects of the launch.
“When only 200 organizations have a capability, it barely moves the needle for the industry. When everyone can point it at their own code and infrastructure, defenders finally get to operate at the same speed as the people attacking them.”
On the cybersecurity implications, Boehm pointed to a long-term rise in disclosed vulnerabilities and expects that trend to continue as AI-assisted software development grows.
“Common vulnerabilities and exposures (CVEs) are up about 7.5x over the last 10 years, from around 6,500 a year to over 48,000. That climb happened before AI was really in the mix. So the question I keep coming back to is whether a release like this causes a short spike in disclosures that then settles, or whether it just keeps going. My honest read is it keeps going, and not because of any single model.
“The amount of software being built — a lot of it now AI-generated — is exploding, and that’s the surface everyone is finding bugs in. Better discovery tools just accelerate a curve that was already steep. So the trend isn’t a blip, it’s a step up on a line that wasn’t slowing down to begin with,” concluded Boehm.
Claude Mythos 5
Alongside Fable 5, Anthropic introduced Claude Mythos 5 for a limited group of cyber defenders and infrastructure providers.
Mythos 5 uses the same underlying model as Fable 5 but operates with safeguards lifted in selected areas. Initial deployments will take place through Project Glasswing, a program run in collaboration with the U.S. government.
The release serves as an upgrade to Claude Mythos Preview.
Earlier this month, the company expanded Project Glasswing from about 50 organizations to roughly 200 partners spanning more than 15 countries.
Pricing and availability
Fable 5 and Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens, less than half the price of Claude Mythos Preview.
The rollout is taking place in stages. Through June 22, Fable 5 is included at no additional cost for Pro, Max, Team, and seat-based Enterprise subscribers. Beginning June 23, access on those plans will require usage credits.
Anthropic said it may extend the included access period if capacity allows and intends to restore Fable 5 as a standard feature of subscription plans when additional capacity becomes available.