Grok 4.1 Thinking Steals #1 Spot on LMArena, Surpassing Google Gemini 2.5 Pro
Do Son 
Elon Musk’s artificial intelligence company, xAI, has launched a surprise offensive with the quiet release of its new Grok 4.1 model series. The update arrives in two variants—the standard Grok 4.1 and the deep-reasoning-enabled Grok 4.1 Thinking—both of which are now freely available to users.
On the LMArena leaderboard, Grok 4.1 Thinking made a dramatic debut at the very top with an Elo score of 1483, while the non-reasoning standard Grok 4.1 swiftly followed, securing second place.
Notably, Google’s previously strong Gemini 2.5 Pro has now slipped to third, trailing the leading Grok 4.1 Thinking by a full 31 points—an unmistakable sign of the pressure mounting ahead of Google’s forthcoming Gemini 3.0 release.
The new models also demonstrate marked improvement in creative writing capabilities. According to Creative Writing v3 benchmark results, both Grok 4.1 Thinking and Grok 4.1 rank just beneath OpenAI’s GPT 5.1, surpassing formidable competitors including OpenAI’s o3, Claude Sonnet 4.5, and Kimi K2 Instruct.
Beyond performance and creativity, xAI has significantly enhanced the models’ accuracy. Data shows that, compared with the previous Grok 4 Fast, Grok 4.1 reduces factual error rates by roughly 70%. Incidents of AI hallucinations have likewise dropped dramatically—from 12.09% to 4.22%—substantially strengthening the system’s practicality and reliability.
Related Posts:
- Cloudflare Unveils AI Crawler Leaderboard: ByteDance Ranks Last
- GPT-4.5 Released: Enhanced Accuracy, Reduced Hallucinations, and Expanded Knowledge
- Privacy Fail: Grok Chatbot Exposes 370,000 Private Conversations
- Grok 2 Goes “Open Source,” But the Catch Is in the Fine Print
Support Our Threat Intelligence
If you find our CVE report and cybersecurity news helpful, consider supporting our work.
Buy Me a Coffee
PayPal
