The AI throne has changed hands overnight!

Early in the morning, Elon Musk, with his company xAI, dropped a bombshell—Grok 4.1 was officially launched and made available to everyone for free.

Interestingly, Grok 4.1 also emphasizes having both high “IQ and EQ,” directly taking on GPT-5.1.

This time, Grok 4.1 comes in two major versions: Grok 4.1 Thinking and Grok 4.1.

On the LMArena leaderboard, Grok 4.1 Thinking scored an impressive 1483 Elo, claiming the title of the world’s top large language model with absolute dominance. The Thinking version is a full 31 points higher than Gemini 2.5 Pro. Even the non-reasoning mode of Grok 4.1 directly surged to second place on the list.

Not only that, Grok 4.1 also boasts exceptional emotional intelligence, with higher emotional awareness, empathy, and interpersonal interaction capabilities. It topped the EQ-Bench with a score of 1586 Elo.

In terms of writing, Grok 4.1 (1722) saw a 600-point Elo increase compared to its predecessor. Moreover, its hallucination rate plummeted by three times compared to previous models.

The rapid evolution of Grok 4.1 can be attributed to the xAI team’s decision to expand the scale of Reinforcement Learning (RL) in the post-training phase by an order of magnitude.

It’s truly unexpected that before Google’s Gemini 3.0 arrives, Musk has made a major move.