On September 1 local time, Anthropic announced the launch of the Claude Fable 5.1 and Claude Mythos 5.1 models. The company called the two models “the world’s most advanced models for coding and knowledge work.”
Fable 5.1 and Mythos 5.1 use the same foundational model, differing in their levels of safety protection. Fable 5.1 is available to all users, while Mythos 5.1 is offered only through a trusted access program to vetted organizations in the cybersecurity and life sciences fields.
In terms of performance, benchmark data disclosed by Anthropic shows that Fable 5.1 outperformed its predecessor Fable 5, Opus 5, and OpenAI’s GPT-5.6 Sol on multiple tests.
On the agentic science research benchmark Terminal-Bench-Science 0.1, Fable 5.1 scored 52.6%, compared with 24.7% for Fable 5, 29.0% for Opus 5, and 22.4% for GPT-5.6 Sol.
On the agentic coding benchmark Terminal-Bench 4.0, Fable 5.1 scored 55.8%, while Mythos 5.1 reached 60.9% with more permissive safety protections.
For knowledge work, Fable 5.1 scored 1853 points on the GDPval-AA v2 test, higher than Opus 5’s 1824 and Fable 5’s 1723.
In terms of pricing, Fable 5.1’s cache read costs have been reduced by 75% to $0.25 per million tokens. Under typical workloads, Fable 5.1 costs about 25% less than Fable 5; for highly agentic workloads, costs can be reduced by up to 45%. Base pricing remains unchanged: $10 per million input tokens and $50 per million output tokens.
For data retention, Anthropic has introduced a new enterprise-grade security measure called Enterprise Frontier Safeguards (EFS). It combines zero-data-retention privacy protection with advanced security safeguards. Customer data is stored on cloud infrastructure fully controlled by the enterprise, rather than in Anthropic’s systems. EFS will be rolled out to enterprise customers in phases beginning this fall. Before EFS becomes available, eligible customers can use Fable 5.1 under the zero-data-retention policy.
Anthropic has improved Fable 5.1’s safety mechanisms to reduce false positives.
In cybersecurity, the updated safeguards produce about 60% fewer false-positive blocks than before. Fable 5.1 can now be used to identify software vulnerabilities, but not to develop exploits targeting those vulnerabilities.
In biology, Anthropic has partnered with the U.S. government to establish an access program allowing researchers to use Mythos 5.1’s advanced biology capabilities, with registration for scientists planned to open soon.
In scientific capabilities, Anthropic demonstrated Mythos 5.1’s achievements across multiple fields. In molecular design, Mythos 5.1 used open-source protein design and folding tools to design high-affinity binders. Across three targets, its binding affinity was 10 times higher than that of the best entries in the Adaptyv Bio protein design competition; its hit rate across 12 targets approached 50%, compared with the current typical hit rate of 10% to 15% for protein design.
In computational analysis, Fable 5.1 trained a neural network based on radar images from NASA’s Magellan mission more than 30 years ago, producing a high-resolution elevation map covering one-third of Venus. Resolution improved from 10 to 20 kilometers to 2 to 3 kilometers, while elevation accuracy increased by 25%.
In computational biology, Mythos 5.1 increased the speed of seven open-source deep learning models by up to 2.5 times by writing custom GPU kernels and caching intermediate results. This is expected to reduce GPU costs by 30% to 60%.
For compliance, Anthropic signed the European Union’s Code of Practice on transparency for AI-generated content under the AI Act in July 2026. Under the act, model outputs released after August 2, 2026, must include an invisible watermark.
Fable 5.1 adds this type of invisible watermark to generated text. (Note: The watermark is not visible to users who do not have access to the detection API. It does not affect output quality or content and contains no information about users or their conversations.)
Anthropic is launching the corresponding detection API as a private preview. Eligible organizations, such as regulators, law enforcement agencies, media organizations, and fact-checking organizations, may use it as required under EU law.
Claude Fable 5.1 is now available on all platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure, and developers can access it through the Claude API. Claude Mythos 5.1 is currently available only to vetted U.S. organizations.