Anthropic released two updated AI models designed to address customer feedback regarding pricing, data handling, and content moderation policies. The new versions, Fable 5.1 and Mythos 5.1, represent iterations on the company’s existing Claude offerings.
The company announced that Fable 5.1 delivers improved performance compared to its predecessor while reducing typical costs by approximately 25 percent. For certain complex agentic applications, savings reach up to 45 percent, largely attributed to reduced pricing on cached data that has been previously processed and stored. Several early access users have provided positive assessments of the model’s capabilities, with evaluations highlighting improvements in coding performance, token efficiency, and more natural communication patterns.
Regarding safety mechanisms, Anthropic stated that Fable 5.1 incorporates more targeted safeguards intended to reduce inappropriate blocks on basic biology-related questions relative to the prior version. Mythos 5.1 maintains the same biology restrictions as its predecessor. The company also detailed progress on data privacy, with Enterprise Frontier Safeguards offering an alternative that stores customer data on their own cloud infrastructure rather than Anthropic’s systems, with rollout beginning later this fall.
In terms of security applications, Anthropic announced that Fable 5.1 can now be utilized for identifying software vulnerabilities, though certain specialized cybersecurity tasks remain routed to other Opus-class models. These restricted applications include penetration testing, exploit development, and binary-based vulnerability analysis. Fable 5.1 is being made available across all platforms, while Mythos 5.1 access is currently limited to Project Glasswing participants.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI