Tech news in 3 minutes
Anthropic’s new Fable release is cheaper, less restrictive
Anthropic released Fable and Mythos 5.1, its most advanced AI models, with reduced token costs, zero data retention for enterprise clients, and new safeguards, setting records in coding and reasoning benchmarks. The Tuesday launch includes performance upgrades and changes to lower false-positive restrictions from safeguards. Fable 5.1, the unrestricted version, is available immediately on cloud platforms or via the Anthropic API. Mythos 5.1 remains limited to registered partners in cybersecurity or life sciences research. A key feature is Anthropic’s zero data retention policy, allowing clients to run models on their own infrastructure without data outflows. A high-privacy service called Enterprise Frontier Safeguards, previously unavailable for Fable due to security concerns, will roll out in the fall. The system monitors for misuse by agents or human users, but clients control how monitoring occurs. Anthropic assured customers that data has not been inappropriately accessed, stating it has never trained on enterprise data without explicit permission. 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 launch, including a custom GPU optimization and a high-resolution map of Venus assembled from existing photos. A detailed system card rates Mythos as “low-risk” for automated AI development concerns, noting its ability to accelerate internal AI R&D is in line with current trends. Mythos 5.1 shows a slight regression in overall misaligned behavior compared to Opus 5 but improves over Mythos 5 and Claude Sonnet 5, being less likely to ignore constraints, hallucinate inputs, or falsely claim task completion.