Anthropic just launched Claude Fable 5, its most powerful Claude model made available to general users. According to Anthropic, Fable 5 reaches Mythos-level capability and shows major improvements across software engineering, knowledge work, vision, scientific research, memory, and long-running complex tasks. The more important part may be how Anthropic is handling access. Claude Fable 5 comes with conservative safeguards, especially around sensitive areas like cybersecurity, biology, chemistry, and model distillation. For certain high-risk requests, users may be routed to Claude Opus 4.8 instead. At the same time, Anthropic is also launching Claude Mythos 5, the same underlying model with fewer restrictions in some areas, reserved initially for trusted cyber defenders and critical infrastructure partners. This is a major signal for the next phase of AI. We are moving from “Who has the smartest model?” to “Who gets access to the most powerful version of intelligence?” That disti...