Grok 4.1 Achieves #1 LMArena Ranking, Quickly Surpassed by Gemini 3

xAI's latest artificial intelligence model, Grok 4.1, briefly claimed the top position on the LMArena Text Arena leaderboard with an impressive 1483 Elo score following its release on November 17, 2025. The model, rolled out to users across grok.com, X, and mobile apps, demonstrated significant advancements in emotional intelligence, creative writing, and a substantial reduction in factual errors. However, its reign was short-lived, as Google's Gemini 3 subsequently launched, immediately surpassing Grok 4.1 with a 1501 Elo score.The rapid shift in leaderboard dominance was highlighted by social media user Dan Loewenherz, who remarked in a tweet, "> Grok 4.1 was the best model for like...3 hours?". This sentiment underscores the intensely competitive and fast-evolving landscape of large language model development. Grok 4.1, which underwent a silent rollout from November 1 to 14, was preferred by users 64.78% of the time over its predecessor in blind evaluations.Grok 4.1 introduces two distinct modes: "Thinking" (codename quasarflux) and "Non-Thinking" (codename tensor). The Thinking mode, designed for complex tasks, secured the top LMArena spot, while the faster Non-Thinking mode ranked second with a 1465 Elo score, outperforming many full-reasoning configurations from rivals. xAI reported a dramatic improvement in emotional intelligence, with Grok 4.1 scoring 1586 on EQ-Bench3, and a 66% reduction in hallucination rates, dropping from 12.09% to 4.22%.The model's creative writing capabilities also saw a substantial boost, achieving a 1708.6 Elo rating on the Creative Writing v3 benchmark, positioning it among the elite AI systems globally. Despite these significant enhancements, the swift succession by Gemini 3 illustrates the relentless pace of innovation in the AI sector, where new benchmarks are set and surpassed within hours or days. This continuous cycle of development challenges companies to maintain leadership amidst fierce competition.