Grok 4.1
Grok 4.1 tops LMArena Text at launch (1483 Elo thinking, 1465 non-thinking) with gains in emotional intelligence and fewer hallucinations.
Post-training focused on style, empathy and fewer factual errors, using a frontier agentic reasoning model as reward model. xAI reports #1 and #2 on LMArena Text and top EQ-Bench3 and Creative Writing v3 scores.
- Date
- Monday, 17 November 2025
- Lab
- xAI
- Kind
- model
- Access
- app only
Figures
| Measure | Value | Measured by |
|---|---|---|
| LMArena Text Elo (thinking) | 1483 #1 at launch; non-thinking 1465 (#2) | company |
Launch ranking is xAI's report of a public leaderboard; leaderboard positions changed within days of other launches.
Sources
This record was checked against its sources on 6 October 2026. How we check