Anthropic Launches Claude Fable 5.1 & Mythos 5.1; Cuts Pricing By Up To 45%

Anthropic has unveiled Claude Fable 5.1 and Claude Mythos 5.1, claiming these are their most sophisticated models for coding and knowledge work. The two models share the same foundation but vary in safety measures. Fable 5.1 is now available to everyone, while Mythos 5.1 is only accessible to trusted users through Anthropic's Cyber Verification Program and Life Sciences Verification Program, the latter created in partnership with the U.S. government.

To tackle customer concerns about costs and privacy, Anthropic is cutting prices by up to 45%. Fable 5.1 will be about 25% cheaper than the previous Fable 5 for usual tasks and could be up to 45% cheaper for complex, multi-step tasks. This price reduction is mainly due to a 75% cut in cache read prices, which are now $0.25 per million tokens. Base prices stay at $10 per million input tokens and $50 per million output tokens.

Anthropic is also introducing Enterprise Frontier Safeguards (EFS), a system that keeps customer data on infrastructure the user controls, not Anthropic's servers, while still preventing misuse. EFS will start rolling out this fall across Claude Code, Claude Enterprise, and major cloud services like AWS, Google Cloud, and Microsoft Azure.

The company highlighted the models' potential in scientific research. Mythos 5.1 reportedly designed high-affinity protein binders with a hit rate near 50% across 12 targets, significantly higher than the usual 10-15%. This result was confirmed by external labs. Meanwhile, Claude Fable 5.1 helped create a high-resolution elevation map of one-third of Venus's surface, using 30-year-old NASA radar data, and will be shared publicly before upcoming NASA and ESA missions.

In terms of safety, Anthropic stated that Mythos 5.1 has stronger cybersecurity capabilities than any previous model, but it remains in the lowest risk category. Additionally, biology-related safeguards now trigger 85% fewer false alarms for harmless medical queries compared to the original safeguards in Fable 5. So far, there's no evidence of a critical security breach for either model.