Anthropic Unveils Claude Fable and Mythos 5.1 for Advanced AI Tasks

Anthropic has announced the release of two new frontier AI models, Claude Fable 5.1 and Claude Mythos 5.1, which are positioned as their most capable systems to date for software development, enterprise knowledge work, and scientific research. While both models share an identical underlying architecture, they are distinguished by the strength of their safety protocols. Fable 5.1 is accessible to all users, whereas Mythos 5.1 is reserved for vetted professionals through Anthropic's trusted access programs, with specific permissions for cybersecurity defenders and life sciences researchers.

The release directly addresses key concerns raised by enterprise customers regarding earlier Claude models: cost, data privacy, and overly stringent content safeguards. Anthropic reports that Fable 5.1 will be approximately 25 percent cheaper than its predecessor, Fable 5, for typical workloads, and potentially up to 45 percent cheaper for heavily agentic tasks. This cost reduction is largely attributed to a significant 75 percent price cut on cached input reads, now priced at $0.25 per million tokens, while input and output token pricing remains unchanged at $10 and $50 per million tokens, respectively.

Data handling has been significantly enhanced through Anthropic's new Enterprise Frontier Safeguards (EFS) framework. EFS allows organizations to store their data within their own cloud infrastructure, offering a zero-data-retention level of privacy while still enabling the detection of adversarial misuse. EFS is scheduled for a phased rollout starting this fall across various Claude platforms and major cloud providers like AWS, Google Cloud, and Microsoft Azure. Until its full deployment, eligible enterprise customers can already utilize Fable 5.1 under a zero-retention policy.

On the safeguards front, Anthropic has observed a 60 percent reduction in false-positive interventions for cybersecurity-related queries. This improvement is partly due to Fable 5.1's new capability to identify software vulnerabilities for defensive purposes, although it remains unable to generate exploits. More sensitive dual-use tasks, such as penetration testing and binary vulnerability scanning, are automatically rerouted to Anthropic's more powerful Opus-class models.

Internal testing by Anthropic indicates that Fable 5.1 surpasses its predecessor across various benchmarks for coding, agentic reasoning, and knowledge work. Early enterprise adopters, including companies like Jane Street, Cognition, MongoDB, Datadog, and Millennium, have reported the model successfully resolving complex issues that had previously stumped engineering teams, including a persistent production crash that had evaded human debugging for nearly five years.

The research capabilities of Mythos 5.1 extend into scientific domains. In molecular biology tests, the model demonstrated success in designing high-affinity protein binders that outperformed leading entries in design competitions, achieving a nearly 50 percent hit rate compared to the industry benchmark of 10-15 percent. Additionally, Fable 5.1 was used to construct a new high-resolution elevation map of Venus using historical NASA Magellan radar data.

Access to Mythos 5.1's advanced biology and cybersecurity features is managed through Anthropic's Cyber Verification Program and Life Sciences Verification Program, developed in collaboration with the U.S. government. Anthropic plans to expand eligibility for these programs over time. The company has also confirmed its compliance with the EU AI Act's Code of Practice, embedding invisible watermarking into outputs from models released after August 2, 2026, and offering a detection API for regulators and researchers.

Claude Fable 5.1 is available immediately through Anthropic's platforms and API under the identifier claude-fable-5-1. This release signifies a step forward in making advanced AI more accessible, cost-effective, and secure for critical enterprise and research applications.