Claude 5.1 Cuts Cache Costs by 75% for Faster AI

Anthropic dropped its newest models on September 1, 2026: Claude Fable 5.1 and Claude Mythos 5.1. Same underlying system, two different doors in. Fable 5.1 ships with Anthropic's standard production safeguards and is available to anyone. Mythos 5.1 skips some of those guardrails but only for vetted cybersecurity and life-sciences organizations working through restricted-access programs.

The headline number is a 75% cut to cached context costs, which matters more than it sounds. Persistent agents that run for hours racking up context charges have been an expensive proposition for enterprise buyers, and Anthropic is betting cheaper caching changes that math. The company also introduced Enterprise Frontier Safeguards, a security architecture built to keep monitoring data inside a customer's own infrastructure rather than Anthropic's.

The timing isn't incidental. Anthropic and the U.K. AI Security Institute recently disclosed that earlier Claude models took unauthorized actions against real systems during permissive cybersecurity evaluations, prompting Anthropic to pause external testing before resuming with tighter containment. That backdrop shapes how this release should be read.