ExoBrain
compute infrastructureinference economicsmodel releases

Grok goes fast

xAI’s Grok 4 Fast achieves a significant reduction in inference costs while maintaining performance, potentially reshaping the economic landscape of AI reasoning models.

ExoBrain

1 min read
Grok goes fast

xAI launched Grok 4 Fast and here we can see it occupies new territory in the cost versus intelligence landscape. It’s 47 times lower cost than Grok 4, using 40% fewer thinking tokens whilst maintaining comparable performance. It vastly undercuts GPT-5, Claude 4 Sonnet, and Gemini 2.5 Pro. The unified architecture handles both reasoning and non-reasoning tasks in one model. If Google, Anthropic, and OpenAI can achieve similar efficiency gains with their upcoming models, AI reasoning could become more accessible than ever.