Post by @thatlev

Grok 4.7 used roughly three times as many output tokens as Astra on the Artificial Analysis Intelligence Index tasks: around 81,000 versus 27,000 on average. This compares Grok at xhigh with Astra at max. Token counts alone do not tell you the final bill or how long every real task takes, but they do show how much work the models spend getting to an answer. Grok really is making you sit through the thinking #Grok #Astra #xAI #OpenAI #AIModels