Anthropic launches Claude Fable 5.1 and says it’s up to 45 percent cheaper for agentic work

Artificial intelligence startup Anthropic has officially introduced its newest generative language models, Claude Fable 5.1 and Mythos 5.1, aimed at resolving persistent enterprise concerns regarding operating costs, safety guardrails, and data privacy.

The newly released Fable 5.1 model boasts enhanced capabilities at a significantly lower price point, offering standard savings of around twenty-five percent and price reductions of up to forty-five percent for intricate, agentic tasks. This cost efficiency is primarily achieved through optimized pricing structures for previously cached data. Prominent technology leaders and early testers have reported that the updated system exhibits remarkable programming aptitude, highly natural conversational speech, and an improved capacity to comprehend subtle data nuances that previous iterations routinely missed.

Alongside these economic and computational enhancements, Anthropic has modified its structural safeguards and information security protocols. The Fable 5.1 model features more precise safety filters designed to prevent the unnecessary blocking of harmless academic inquiries, and it is now approved for identifying vulnerabilities in software code. However, highly complex cybersecurity operations, such as penetration testing and exploit generation, are still routed to the company's larger Opus models. To satisfy strict corporate privacy requirements, Anthropic plans to deploy new enterprise safeguards later this autumn, allowing clients to store their proprietary data on their own private cloud infrastructure instead of Anthropic's host servers. Claude Fable 5.1 is now accessible across all consumer channels, whereas Mythos 5.1 remains restricted to select test participants.