Elon Musk-owned xAI has started rolling out Grok 4.1, which is an upgrade to the existing Grok 4 model.

xAI has launched two Grok 4.1 models with some incremental improvements.

One of the models is Grok 4.1, and another is Grok 4.1 Thinking. These two models are available for free, but paying customers get fewer limits.

According to xAI, Grok 4.1 is 3x less likely to hallucinate compared to its previous models, which makes it one of the best releases by xAI.

We don't know how well it performs against its rivals, such as GPT 5.1, which recently shipped with performance and emotional intelligence improvements.

However, LMArena's Text Arena shared some interesting performance insights of Grok 4.1

LMArena's Text Arena is an open-source tool that allows users to compare different large language models (LLMs) through side-by-side, blind, and randomised tests.

According to early benchmarks, Grok 4.1 (thinking) and Grok 4.1 have scaled new heights in the most competitive Text Arena.

According to benchmarks, Grok 4.1 (thinking) also ranks at #1 with a score of 1510 and Grok 4.1 ranks at #19 with a score of 1437 in the Arena Expert leaderboard.

"This is a 40+ point improvement since Grok 4 fast, which landed in the Arena just two months prior," the benchmark platform noted.

While Grok 4.1 is a decent upgrade, it might not be the best model of this year, as Google is preparing Gemini 3.0, which could be the most powerful model till date.