Anthropic Introduces Claude Fable 5.1 and Mythos 5.1 Models

Anthropic has introduced the Claude Fable 5.1 and Claude Mythos 5.1 models, designed for programming and scientific research.

The company stated that the Fable 5.1 model has been released for public use. Mythos 5.1 is being provided with a special security system for trusted partners in the fields of cybersecurity and biology.

The new Fable 5.1 is up to 25 percent cheaper than the previous version. Cost savings for agentic tasks are reported to reach up to 45 percent.

Anthropic has also implemented the Enterprise Frontier Safeguards system, which allows users to store their data entirely within their own cloud environment.

Fable 5.1 has recorded high results in a series of benchmark tests related to complex problem-solving and scientific research. In the Terminal-Bench-Science test, the model achieved a score of 52.6 percent.

In the Terminal-Bench 4.0 test, specialized for programming, Fable 5.1 achieved 55.8 percent accuracy. The company emphasizes that these results demonstrate the model's high efficiency in domain-specific tasks.