Anthropic ships Claude Fable 5.1

Anthropic released Claude Fable 5.1 on September 1 as a generally available model for coding and knowledge-work tasks, while reserving Claude Mythos 5.1 for customers in its trusted-access programs. The company said the two offerings are the same underlying model, but with different safeguard settings and access rules. (anthropic.com) Anthropic paired the launch with new enterprise controls on data retention and monitoring after business customers pushed back on its earlier requirement to keep logs for 30 days. The split matters because Anthropic is now selling not just model capability, but the terms under which companies can use that capability. (anthropic.com) Fable 5.1 is positioned for broad enterprise deployment. Mythos 5.1 is being held back for approved cybersecurity and life-sciences users, with more permissive safeguards for that work. ### Why did Anthropic ship two versions of the same model? Anthropic said Claude Fable 5.1 and Claude Mythos 5.1 are “the same model,” with the difference lying in the safeguard layer wrapped around them. Fable 5.1 is generally available. Mythos 5.1 is limited to vetted users through Anthropic’s trusted-access programs and is designed for cybersecurity and life-sciences use cases that the company considers more sensitive. (anthropic.com) Anthropic’s Mythos product page says Fable 5.1 now allows some vulnerability-identification work in source code, while still blocking penetration testing, exploit generation and binary-based vulnerability scanning. (anthropic.com) The company also said its updated biology safeguards intervene on benign requests 85% less often than the safeguards it launched with Fable 5. ### What changed on data retention after customer complaints? Anthropic said Fable 5.1 addresses customer feedback on price, data retention and safeguards. CNBC reported on September 1 that the company was changing a controversial data-retention policy for business customers after receiving “a lot of feedback.” Anthropic had tied the earlier 30-day retention rule to its covered-model rollout, arguing the logs were needed to detect sophisticated misuse across sessions and accounts. (anthropic.com) Enterprise Frontier Safeguards is Anthropic’s answer to that dispute. Anthropic and partner accounts said the system lets eligible customers use covered models while keeping activity data in cloud environments they control, rather than handing routine log custody to Anthropic. (anthropic.com) Help Net Security reported that the setup keeps Claude logs in a customer’s own cloud under its keys while Anthropic’s systems still monitor for misuse. ### Where is Fable 5.1 actually available? Anthropic said Fable 5.1 is available through Claude, the Claude API and major cloud distribution channels. The company’s Fable page says the model is available to Pro, Max, Team and Enterprise users, while developers can access it on the Claude Platform and through Amazon Web Services, Google Cloud and Microsoft Foundry. (anthropic.com) AWS separately confirmed availability through Amazon Bedrock and Claude Platform on AWS. Pricing was part of the launch. Anthropic said Fable 5.1 takes steps to address customer concerns on price, and outside coverage reported a 75% cut in cache-read pricing for Fable 5.1 compared with Fable 5. (anthropic.com) VentureBeat reported cache-hit input pricing fell to $0.25 from $1.00. ### What does the enterprise emphasis tell employers and engineers? Anthropic’s launch materials and partner briefings focused on guarded deployment, customer-held logs and access controls for more capable models. That framing suggests companies adopting these systems are being asked to prove not only what the model can do, but how its use is constrained. (anthropic.com) For junior engineers, that means interview questions are likely to move beyond whether they use coding assistants at all. Employers rolling out tools with retention settings, policy controls and restricted-access tiers are more likely to ask how generated code was tested, what data was exposed to the model, and which tasks should stay off-limits. (anthropic.com) That is an inference from Anthropic’s product design and enterprise controls, not a statement the company made directly. ### What comes next for the rollout? Anthropic said Enterprise Frontier Safeguards will be supported on Claude Code, Claude Enterprise, the Claude Platform, Amazon Bedrock, Claude Platform on AWS, Google’s Agent Platform and Microsoft Foundry. (anthropic.com) Becker’s Hospital Review and Anthropic partner materials said the broader safeguards rollout is planned for the fall, while Mythos 5.1 remains limited to approved organizations in Anthropic’s trusted-access programs. (anthropic.com)