Anthropic Launches Claude Fable 5.1, Cutting Agentic AI Costs by Up to 45%
Updated
Updated · The Verge · Sep 1
Anthropic Launches Claude Fable 5.1, Cutting Agentic AI Costs by Up to 45%
3 articles · Updated · The Verge · Sep 1
Summary
Claude Fable 5.1 is now live across Anthropic’s platforms, with the company saying it outperforms Fable 5 while cutting typical costs about 25% and complex agentic-task costs by as much as 45%.
Cached-data pricing drives much of the savings, while Anthropic also says the update answers customer complaints about overly aggressive safeguards and data handling.
Fable 5.1’s safeguards are designed to reduce false blocks on basic biology questions, and Anthropic is now permitting it to help identify software vulnerabilities, though higher-risk cybersecurity work still gets routed to Opus models.
Enterprise Frontier Safeguards will begin rolling out later this fall, letting customers keep data on their own cloud servers rather than Anthropic’s; Mythos 5.1 is available only to Project Glasswing participants.