Anthropic has officially released Claude Mythos 5, its most advanced AI model, alongside Claude Fable 5—a publicly accessible version equipped with stringent safeguards against misuse. Mythos 5 is restricted to approved cybersecurity firms, critical infrastructure operators, government partners, and select life sciences researchers due to its demonstrated ability to autonomously discover software vulnerabilities and execute complex cyberattack simulations. The model previously made headlines when a preview version completed a 32-step corporate network intrusion exercise without human assistance and found over 271 vulnerabilities in Mozilla Firefox.
Fable 5, built on the same core architecture as Mythos 5, is now available to the general public through Anthropic’s Claude API and enterprise plans. It excels at software engineering, knowledge work, and vision-based tasks, achieving a 90% score on Hex’s core analytics benchmark and outperforming competitors in UI design and game coding. However, to prevent misuse, Fable 5 automatically delegates requests involving cybersecurity, biology, chemistry, and model distillation to the less powerful Claude Opus 4.8—a fallback triggered in fewer than 5% of sessions. Anthropic says the model has undergone rigorous red-teaming, with over 1,000 hours of external bug bounty testing yielding no universal jailbreaks.
The launch includes a mandatory 30-day data retention policy for all traffic, even for enterprises with prior zero-retention agreements. Anthropic states the data will not be used for training but solely to “defend against complex and novel attacks” and reduce false positives. Pricing for both models is set at $10 per million input tokens and $50 per million output tokens—double the cost of Opus 4.8. Subscription users can access Fable 5 at no extra cost through June 22, after which it will require usage credits until restored as a standard feature.
Anthropic’s move comes amid easing tensions with the Trump administration and as the company prepares for a public market debut alongside OpenAI and SpaceX. The rollout also follows Anthropic’s call for a global “brake pedal” on frontier AI development, warning that systems may soon achieve recursive self-improvement. While Mythos 5 is positioned as a cybersecurity defense tool, critics, including OpenAI’s Sam Altman, have accused the firm of fear-based marketing.
