SAN FRANCISCO — Anthropic announced the launch of two new AI models, Claude Fable 5.1 and Claude Mythos 5.1, positioning them as the company’s most advanced offerings for coding and knowledge-work tasks. Both models share the same underlying architecture, but they diverge in the level of safeguards applied. Claude Fable 5.1 is generally available, while Claude Mythos 5.1 is available only through trusted access programs and carries safeguards specifically designed to support work in cybersecurity and the life sciences.
Anthropic highlighted a suite of improvements that respond to customer feedback on price, data-retention policies, and safeguards. For typical token-based workloads, Fable 5.1 is priced about 25 percent lower than its predecessor, Fable 5, due to reduced pricing on cache reads.
In highly agentic work, savings can be up to approximately 45 percent.
Enhanced security and privacy for enterprises
The rollout introduces Enterprise Frontier Safeguards (EFS), a privacy framework that offers complete privacy equivalent to a zero data retention policy while remaining state-of-the-art at preventing adversarial use. EFS stores data in cloud infrastructure controlled entirely by the customer, not Anthropic, and will be made available to enterprise customers in phases beginning later in the fall. Until EFS is available, eligible customers can use Fable 5.1 with zero data retention.
Safeguards have been improved to reduce false positives; in cybersecurity, the newest safeguards block 60 percent fewer false positives than before.
Capabilities for developers and scientists
Claude Fable 5.1 can now be used to discover software vulnerabilities, though not to develop exploits for them. In biology, an access program developed in partnership with the US government will enable access to Claude Mythos 5.1’s advanced biology capabilities, with enrollment for scientists expected to open soon. Performance charts released with the announcement show Claude Fable 5.1 surpassing the older Fable 5 across a range of benchmarks, delivering results similar to or better than Fable 5 at a much lower cost when operating at Low or Medium effort levels.
The model defaults to High effort in Claude Code, Medium in Claude Cowork, and Medium on Claude.ai. Benchmark scores included Terminal-Bench-Science 0.1, Terminal-Bench 4.0, Humanity’s Last Exam, and CursorBench 3.2.0, with cost per task comparisons across effort levels.
The performance gap between Fable 5.1 and Mythos 5.1 on certain tasks reflects earlier, less precise cyber safeguards. Anthropic expects that with improvements, the difference will be much smaller.
What to watch next
Stakeholders will be monitoring the phased deployment of Enterprise Frontier Safeguards later in the fall, as well as the opening of the biology access program for qualified scientists. The combination of lower pricing, stronger privacy guarantees, and expanded security capabilities suggests that both developers and researchers could adopt these models more broadly in the months ahead.
