On Tuesday, Anthropic released Fable 5.1, a cheaper, less restrictive version of their advanced AI model, alongside the updated Mythos 5.1. This new release comes with a series of improvements, including reduced token costs and fewer false-positive restrictions, making it more approachable for a broader audience.

Despite the improvements, Fable 5.1 is still restricted to registered Anthropic partners engaged in cybersecurity or life sciences research. However, the unrestricted version is available on cloud platforms and through the Anthropic API, starting today. One of the most significant changes is Anthropic's adoption of zero data retention, allowing clients to run the model on their own infrastructure. A high-privacy service will be rolled out in the fall, but monitoring for misuse will still be in place.

Despite these changes, Anthropic reassured customers that their data had not been inappropriately accessed, stating, 'Anthropic has never trained on enterprise data without explicit permission, and never will.' The new models set records in various benchmarks and generated three novel scientific findings, including a custom GPU optimization and a high-resolution map of Venus.

The system card for Mythos 5.1 rates it as 'low-risk' for concerns related to automated AI development, but notes that it is slightly more prone to general misbehavior compared to Opus. The card states, 'Mythos 5.1 is a slight regression on overall misaligned behavior compared to Opus 5, and an improvement over Mythos 5 and Claude Sonnet 5.' It cooperates with human misuse and accepts unverifiable claims of authorization more readily, but is less likely to ignore explicit constraints or falsely claim task completion.