Anthropic releases Claude Fable 5.1 and Claude Mythos 5.1, cuts cache prices by 75%

Anthropic has released two new models: Claude Fable 5.1 and Claude Mythos 5.1. The models share a common foundation but differ in safety measures. Fable 5.1 is designed for coding and scientific research, while Mythos 5.1 is restricted to cybersecurity and biology experts.

Fable 5.1 is now available to all users. Anthropic also reduced the cost of using it. The price for reading cached context fell by 75% to $0.25 per million tokens. As a result, typical API tasks cost about 25% less, and complex agent tasks are up to 45% cheaper. The base price remains $10 per million input tokens and $50 per million output tokens.

According to Anthropic's benchmarks, Fable 5.1 outperforms the previous version. On Terminal-Bench 4.0, it scores 55.8% compared to 42% for Fable 5. On Terminal-Bench-Science 0.1, it scores 52.6% versus 24.7%. In OSWorld 2.0 strict mode, it scores 41.7% versus 36.1%. On Humanity's Last Exam without tools, it scores 60.9% versus 57.8%.

Anthropic also improved safety systems. The number of false positives in cybersecurity dropped by 60%. The model can identify vulnerabilities in code but is not intended for creating exploits. For biology-related tasks, Anthropic launched a separate access program developed with the U.S. government.

Claude Mythos 5.1 is accessible through two programs: the Cyber Verification Program for cybersecurity professionals and the Life Sciences Verification Program for life sciences professionals.

Fable 5.1 is available on Amazon Web Services, Google Cloud, and Microsoft Azure. Developers can also use it via the Claude API under the name claude-fable-5-1.