Grok 4.5 Tops Agent Test, Backing Musk’s Opus-Class Claim

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 (48.6%) and Claude Opus 4.8 (48.5%). An The post Grok 4.5 Tops Agent Test, Backing Musk’s