Elon Musk called Grok 4.5 an Opus-class model that runs faster and costs less. An independent, agentic benchmark now provides real backing for that claim. On Artificial Analysis’s AutomationBench-AA, Grok 4.5 ranked first with a 51.4% score while costing $0.34 per task. It beat both Claude Fable 5 (