A version of an artificial intelligence (AI) system that its maker previously said was too powerful for broad release has been launched to the public with new safeguards.
Anthropic on Tuesday introduced Claude Fable 5, describing it as a “Mythos-class” model designed for general use, alongside Claude Mythos 5, a more permissive version intended for a limited group of cybersecurity professionals and infrastructure providers.
Mythos, launched in April, was initially restricted to a small group of tech companies due to cybersecurity concerns. Reports said it is not only effective at identifying system vulnerabilities but also at linking them into complex exploit chains, making it a powerful tool for hacking.
“Fable 5’s capabilities exceed those of any model we’ve ever made generally available,” the company said, adding that it shows state-of-the-art performance across software engineering, knowledge work, vision and scientific research benchmarks.
According to Anthropic, Claude Fable 5 includes safeguards that redirect requests related to hacking, biology, and chemistry to Claude Opus 4.8, a less capable model, to help prevent misuse.
Claude Mythos 5 succeeds the Mythos Preview, which was previously released to a limited group of organizations under Project Glasswing for testing. According to Anthropic, during that phase, users were able to identify more than 10,000 critical security flaws in their systems using Mythos Preview.
Anthropic said Mythos 5 has “the strongest cybersecurity capabilities of any model in the world.”
Like its predecessor, Mythos 5 will first be made available to a limited group of technology companies under Project Glasswing.
Anthropic said the models are already showing potential benefits in cybersecurity defense and scientific research, including helping identify vulnerabilities and supporting drug discovery efforts.
“We’ve seen the beginnings of this in Project Glasswing, where the models have helped cyber defenders secure critically important software. We’ve also seen it in life sciences research, where the models are positing novel hypotheses and speeding up the development of new therapeutics,” Anthropic said.
Pricing for both Claude Fable 5 and Claude Mythos 5 has been set at $10 per million input tokens and $50 per million output tokens, less than half the cost of the prior Mythos Preview system.
