Why Anthropic is 'refusing' to release AI model that it says is most powerful it made

Anthropic has confirmed it is testing Claude Mythos, a new AI model it describes as the most capable it has ever built—and the most dangerous to release. The model introduces a new tier above Opus called Capybara, and significantly outperforms Claude Opus 4.6 in coding, reasoning, and cybersecurity. Anthropic isn't rushing a wider rollout because, by its own assessment, Mythos is currently far ahead of every other AI model in cyber capabilities.