Anthropic Fable Mythos Relaunch

Anthropic relaunched its flagship line, and the headline isn't intelligence, it's economics.

Claude Fable 5.1 and Mythos 5.1 replace Fable 5 at up to 45 percent lower cost for heavily agentic workloads, driven mostly by a 75 percent cut to cache-read pricing. Typical workloads get a 25 percent discount alongside what Anthropic says is meaningfully better performance.

That's the real story for anyone running AI at scale this year. The frontier labs stopped competing only on benchmarks and started competing on the unit economics of running an agent continuously, rather than answering a single prompt. Cache-read pricing sounds like a technical footnote, but for a team running long-context agents that call tools and hold context across a workday, it's the line item that decides whether a workflow is actually profitable.

Anthropic also split the release in two: Fable 5.1 is generally available, while Mythos 5.1 stays gated to vetted researchers and defenders working in cybersecurity and life sciences. That's a deliberate signal about where the highest-capability models are heading, and who gets access first.

If your team is budgeting AI spend for next quarter, the model card matters less than the token math behind it.

What would a 25 to 45 percent cut in your AI compute bill change about which workflows you're willing to automate?