Anthropic releases Claude Mythos preview
Anthropic’s most powerful reasoning model, Claude Mythos, is transitioning from a restricted security preview to a public release via Claude Code and Claude Security. The "step-change" model enables autonomous vulnerability discovery and complex codebase refactoring.
Mythos is Anthropic's "nuclear option" for code—a model so powerful it was initially restricted to elite defensive partners under Project Glasswing.
- –Transitions from restricted access to a commercial preview labeled claude-mythos-1-preview.
- –Integrated directly into the Claude Code CLI, moving beyond simple completions to reasoning-heavy autonomous development.
- –Claude Security uses Mythos to automate patch generation, shifting from mere vulnerability flagging to proactive remediation.
- –Staged rollout demonstrates a new "safety-first" framework for deploying dual-use frontier models in commercial environments.
DISCOVERED
5h ago
2026-05-26
PUBLISHED
9h ago
2026-05-26
RELEVANCE
AUTHOR
The_Cyber_News