Anthropic Releases Claude Mythos Preview with Cybersecurity Capabilities but Withholds Public Access

Anthropic has released Claude Mythos Preview, its most capable AI model to date, featuring dramatic improvements in reasoning, coding, and cybersecurity. Unlike previous releases, it will not be publicly available. Instead, access is restricted to a consortium of major tech companies — including AWS, Apple, Google, Microsoft, and others — through a new initiative called Project Glasswing, backed by $100M in usage credits. Internal benchmarks show the model autonomously discovered and exploited zero-day vulnerabilities across major operating systems and browsers, including a 27-year-old OpenBSD bug and a 16-year-old FFmpeg vulnerability. SWE-bench verified scores reached 93.9%, up from 80.8% for Opus 4.6. Community reaction is mixed, with some raising concerns about embedded devices that can't be patched, and others questioning whether benchmark performance translates to broader usefulness or cost-effectiveness.