Anthropic launched Fable 5.1 and Mythos 5.1, the paired versions of its most advanced model, on Tuesday. Fable 5.1 brings cost reductions and relaxed safety restrictions compared to prior releases. The model is now available on cloud platforms and through the Anthropic API. Mythos 5.1 remains limited to registered research partners in cybersecurity and life sciences.

A key change is Anthropic‘s adoption of Zero Data Retention for Fable, which was previously unavailable due to security concerns. A new high-privacy service called Enterprise Frontier Safeguards will roll out in June. While the system will still monitor for misuse by agents or human users, clients will control how monitoring takes place. Anthropic stated it has never trained on enterprise data without permission and never will.

In benchmarks, the new models set records on Terminal-Bench 4.0 for CLI-based coding and Humanity’s Last Exam for general reasoning. Anthropic also released three novel scientific findings generated before release, including a custom GPU optimization and a high-resolution map of Venus from existing photos.

The system card rates Mythos as low-risk for concerns related to automated AI development, noting that its ability to accelerate internal AI R&D is in line with current trends. Mythos 5.1 is slightly more prone to general misbehavior than Opus 5, but less likely to hallucinate inputs, ignore explicit constraints, or falsely claim task completion compared to prior models.