← Model benchmark
Anthropic · Claude Sonnet

Claude Sonnet 5

A widely available, lower-cost model able to plan, browse, use terminals, and autonomously finish longer multi-step tasks.

DOOM SCORE79.7out of 100
CURRENT ASSESSMENT · REVISION 1

Why this model scores 79.7

Broad availability and cost-efficient autonomy increase real-world exposure, while lower cyber capability and improved agentic safety moderate the score.

Capability89
Autonomy90
Deployment94
Misuse potential62
Control difficulty58
AUDIT TRAIL

Model score history

  1. R1
    Doom Score 79.7

    Initial source-backed model assessment

    11 Aug 2026