xAI announced its new model called Grok 4 Fast. Grok 4 stands out with its fast and efficient operation; while maintaining the same performance level as the current Grok 4, it offers an average of 40% savings in token usage. Its faster and cheaper approach significantly reduces costs while improving user experience. Data shared by the company shows that Grok 4 Fast can reduce the cost of the same performance by 98%. This provides users with an advantageous experience in tasks like coding or quick web browsing.
One of the most notable features is its flexible architecture. Similar to the system used in OpenAI’s GPT-5, Grok 4 Fast integrates two different approaches:
- Reasoning model: activates for tasks requiring complex and deep thinking.
- Non-reasoning model: used for quick and superficial responses.
This dual structure offers users the flexibility to receive more accurate or faster answers according to their needs.
Achievements from testing: On the independent comparison platform LMArena, Grok 4 Fast ranked first in search-focused tasks and eighth in text-focused tasks.
User access: xAI not only offers Grok 4 Fast to paying users but also makes it freely accessible to all users via web, iOS, and Android. However, in a field where competition is rapidly intensifying, it is expected that competitors like Google’s new Gemini model or updates from Anthropic’s Claude Opus 4.1 will arrive soon.