🕒 Created · Updated

Anthropic releases Fable 5.1 and Mythos 5.1 AI models to improve coding and science performance at lower costs

Anthropic PBC has launched the latest versions of its flagship AI models, Fable 5.1 and Mythos 5.1. The administration announced that these new models offer stronger performance in coding and scientific tasks while reducing costs for users. Specifically, the company reduced the price of cached context by 75 percent, which lowers the total cost of running complex, long-running tasks. Fable 5.1 is generally available, while Mythos 5.1 remains restricted to a trusted access program for organizations in cybersecurity and life sciences. Both models are based on the same underlying architecture but feature different safeguards. Fable 5.1 includes more precise production safeguards that reduce interventions by approximately 60 percent in coding sessions. To address enterprise needs, the company is introducing Enterprise Frontier Safeguards, which allows customers to store monitoring data in their own cloud infrastructure. This system provides the privacy of zero data retention while allowing Anthropic to monitor for misuse. The new models also feature improved safety classifiers to better handle biology and cybersecurity tasks without blocking benign queries.

Sources

Paywall and unreadable sources