Anthropic releases public Mythos-class Claude Fable 5

Anthropic has released Claude Fable 5, its first Mythos-class model available to the public. The company previously said its Claude Mythos model was too powerful for general release and instead shared access with software vendors including Apple for cybersecurity applications.

Anthropic positions Fable 5 as state-of-the-art. The company says Fable 5’s capabilities exceed those of any model it has ever made generally available. It is state-of-the-art on nearly all tested benchmarks of AI capability, showing exceptional performance in software engineering, knowledge work, vision, scientific research, and many other areas.

Anthropic adds that the longer and more complex the task, the larger Fable 5’s lead over its other models.

Safeguards limit certain queries on Fable 5. Releasing a model this capable comes with risks, Anthropic states. Without safeguards, Fable 5’s capabilities in areas like cybersecurity could be misused to cause serious damage.

The company has therefore launched the model with safeguards that mean queries on some topics will instead receive a response from its next-most-capable model, Claude Opus 4.8. These safeguards are tuned conservatively and will sometimes catch harmless requests, though they trigger on average in less than 5% of sessions.

Anthropic plans to refine the safeguards. With more capable models arriving in the coming months, the company is working to improve its safeguards and reduce false positives as quickly as it can.

Mythos 5 remains restricted. Beyond Claude Fable 5, Anthropic also details Claude Mythos 5, which is basically Fable 5 without a lot of the safeguards. Mythos 5 is only available to a small group of cyberdefenders and infrastructure providers.