← Model benchmark
Anthropic · Claude Mythos

Claude Mythos 5

Anthropic's most capable cyber and biology research model, initially limited to vetted users and high-security deployments.

DOOM SCORE81.7out of 100
CURRENT ASSESSMENT · REVISION 1

Why this model scores 81.7

Extreme dangerous-domain capability and difficult residual control produce a high score, but tightly restricted access substantially lowers current deployment exposure.

Capability96
Autonomy92
Deployment22
Misuse potential96
Control difficulty84
AUDIT TRAIL

Model score history

  1. R1
    Doom Score 81.7

    Initial source-backed model assessment

    11 Aug 2026