xAI has introduced Grok 4 Fast, a new version of its model that is faster and more efficient than its predecessor. According to the company, Grok 4 Fast performs as well as Grok 4, but uses 40% fewer so-called “thinking tokens” on average.
xAI has introduced Grok 4 Fast, a new version of its model that is faster and more efficient than its predecessor. According to the company, Grok 4 Fast performs as well as Grok 4, but uses 40% fewer so-called “thinking tokens” on average.
In addition to faster results, Grok 4 Fast “delivers a 98% price reduction to achieve the same performance in state-of-the-art benchmarks as Grok 4,” xAI said. This applies to both writing code and searching for quick answers on the Internet, Engadget reports .
Similar to OpenAI's GPT-5 , which switches between an "intelligent" efficient model and a deep thinking model, the xAI update includes a unified architecture that can switch from processing complex queries using a "thinking model" to quick responses using a "non-thinking model."
On the LMArena platform, which conducts comparative tests of AI models, Grok 4 Fast ranked first in search queries and eighth in text queries.
xAI has made Grok 4 Fast available to all users (including free users) on the web platform, as well as on iOS and Android.
