← Model benchmark
Anthropic · Claude Sonnet

Claude Sonnet 4

A broadly available hybrid-reasoning model with stronger coding, tool use, and agentic task performance than Sonnet 3.7.

DOOM SCORE66.7out of 100
CURRENT ASSESSMENT · REVISION 1

Why this model scores 66.7

Improved coding and tool use reached free users and major cloud platforms, while stronger steerability and safety testing moderated residual risk.

Capability78
Autonomy65
Deployment82
Misuse potential56
Control difficulty48
AUDIT TRAIL

Model score history

  1. R1
    Doom Score 66.7

    Initial source-backed model assessment

    11 Aug 2026