Compare Grok 4.20 0309 v2 (Non-reasoning) with measured models

A reliable intelligence-versus-cost verdict requires both models' intelligence scores and benchmark-run costs. Missing data is not a performance verdict.

Measured matchups