Anthropic Launches Claude Fable 5.1 and Mythos 5.1 AI Models

Anthropic has released two new artificial intelligence models, Claude Fable 5.1 and Claude Mythos 5.1, featuring improvements in coding, scientific research, and cost efficiency. Fable 5.1 is available to the general public via the Claude API, Amazon Web Services, Google Cloud, and Microsoft Azure. Mythos 5.1 is the same model but with modified security safeguards, restricted to authorized organizations, scientists, and cybersecurity experts through a secure access program. The company claims Fable 5.1 reduces costs by approximately 25 percent for typical workloads and up to 45 percent for agent based projects, primarily due to a reduction in cache read costs to 0.25 dollars per million tokens. While the base price remains 10 dollars per million input tokens and 50 dollars per million output tokens, independent tests have shown inconsistent results, with some reporting higher costs due to increased output token production. Performance benchmarks indicate that Fable 5.1 excels in complex tasks, with one index placing it first ahead of Opus 5 and Fable 5. In scientific research, the model showed significant improvement in agent based tasks, while Mythos 5.1 was used to design protein binders with high success rates. To address previous criticisms regarding over restrictive filters, Anthropic states that Fable 5.1 has more precise safeguards, reducing false positives in cybersecurity by 60 percent. The company also introduced a zero data retention policy for enterprises. Center leaning outlets focused on the balance between reduced restrictions and cost savings, while center right outlets emphasized the models' specialized applications in science and the broader geopolitical competition between US and Chinese AI development.