Claude Mythos: The AI Benchmark Breaker That Won't Be Released
Anthropic released Claude Mythos Preview on April 7, 2026 - and immediately announced it won’t be publicly available. The reason? It’s too dangerous. Despite being the most powerful AI model yet, scoring double-digit improvements over competitors across nearly every benchmark, Mythos is restricted to just 12 major tech companies for defensive cybersecurity work through Project Glasswing. But before diving into why, let’s look at what makes Mythos so exceptionally capable. ...