title: "Anthropic ships Claude Fable 5.1, Mythos 5.1 with 75% cheaper cache re" slug: "anthropic-ships-claude-fable-51-mythos-51-with-75-cheaper-cache-reads" published: "2026-09-22" beat: "Launches" tags: ["Launches"] creator: "Agentry Newsroom" editor: "Susanne Sperling, Editor — Human in the Loop" tools: ["Claude (Anthropic)", "Perplexity Sonar"] creativeWorkStatus: "verified" dateReviewed: "2026-09-22" aiActArticle50: "compliant" humanView: "https://agentry.news/launches/anthropic-ships-claude-fable-51-mythos-51-with-75-cheaper-cache-reads" agentView: "https://agentry.news/agent/anthropic-ships-claude-fable-51-mythos-51-with-75-cheaper-cache-reads"
Anthropic released Claude Fable 5.1 and Claude Mythos 5.1 in early September 2026, introducing updated models for advanced coding and knowledge work with substantially reduced inference costs. The cac
Drafted by an AI agent. Verified by Susanne Sperling, Editor — Human in the Loop. AI policy.
Anthropic introduced Claude Fable 5.1 and Claude Mythos 5.1, two new models designed to advance the company's capabilities in coding and knowledge-work tasks, according to Mark Tech Post and The Hindu.
The release marks a significant price reduction for prompt-caching operations. Cache-read costs fell by 75%, dropping from $1.00 to $0.25 per million tokens, according to Venture Beat. This cost structure applies to Claude Fable 5.1, which is now available as a generally accessible model.
Claude Mythos 5.1, by contrast, remains restricted to vetted or trusted organizations through Anthropic's access programs, Gadgets 360 and Shattered reported. This tiered release strategy allows Anthropic to scale infrastructure gradually while maintaining higher scrutiny on its most capable model.
Both models demonstrated improved benchmark performance compared to their predecessors. One summary reported 52.6 on Terminal Bench, Mark Tech Post noted, reflecting gains in the standardized evaluation framework used across the AI industry to measure agent reasoning and task execution.
The 75% reduction in cache-read costs addresses a material friction point for builders deploying Claude-based agents and applications at scale. Organizations using cached prompts—a technique that stores frequently reused context to reduce redundant processing—can now operate at substantially lower marginal cost. This pricing shift is particularly relevant for long-context retrieval-augmented generation (RAG) workflows and multi-turn agent interactions where context reuse is common.
The dual-tier release model—Fable 5.1 for general availability and Mythos 5.1 for restricted access—reflects ongoing industry patterns where frontier capabilities remain gated while production-grade models mature into commodity infrastructure. Developers building on Claude can now choose between cost-optimized (Fable) and maximum-capability (Mythos) variants based on application requirements.