AI BriefingGrok 4 Fast
Key point
xAI has unveiled Grok 4 Fast, which combines excellent cost efficiency with strong reasoning performance.
Details
xAI announced Grok 4 Fast, a cost-efficient reasoning model. Built on the training experience from Grok 4, this model delivers state-of-the-art performance in both enterprise and consumer domains, and boasts outstanding token efficiency.
It features a 2M token context window and a single unified architecture that integrates reasoning and non-reasoning modes. Through large-scale reinforcement learning (RL), it maximized intelligence density, maintaining performance similar to Grok 4 while cutting average thinking token usage by 40%.
It is also innovative in terms of cost. It achieved a 98% reduction in the cost of delivering the same performance as Grok 4, and recorded a SOTA-level price-to-intelligence ratio on Artificial Analysis's Intelligence Index.
Key capabilities include the following:
- Tool-use reinforcement learning: excels at tool calling capabilities such as code execution and web browsing.
- Agentic search: explores the web and X in real time, rapidly synthesizing multimedia information such as images and videos.
- LMArena results:
grok-4-fast-searchtook first place in the Search Arena, and also showed overwhelming performance among comparable models in the Text Arena.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.