analiza aiqbench

Anthropic is launching Claude Fable 5, its first publicly accessible Mythos-class model, which delivers top-tier capabilities in software engineering and vision but enforces hard safety guardrails on high-risk domains like cybersecurity and biology. While the model is temporarily included in existing subscription tiers, Anthropic plans to transition it to a usage-credit model by June 23, reflecting a strategic shift toward monetizing frontier AI while maintaining strict access controls. This release underscores the industry's balancing act between democratizing advanced model access and mitigating risks associated with rapid autonomous self-improvement.