Grok 4.1

Grok 4.1 is xAI's November 2025 flagship, focused on conversational quality: faster reasoning, improved emotional intelligence and style, and significantly reduced hallucination rates. It rolled out to all consumer users on grok.com, X and the mobile apps on day one, and briefly topped the LMArena Text leaderboard (Grok 4.1 Thinking ~1483 Elo) before Gemini 3 Pro took the lead. Two days later xAI shipped Grok 4.1 Fast — a cheaper API variant with a 2M-token context window that xAI calls its best tool-calling model — plus an Agent Tools API. The API launched at $0.20/$0.50 per 1M input/output tokens, among the cheapest frontier options at the time. Closed weights; access is via xAI's API or the consumer apps.

Provider-reported (xAI white paper / announcements): LMArena Text ~1483 Elo (Thinking variant; ~1465 standard) · EQ-Bench 3 ~1586 Elo (near top among published systems) · Creative Writing v3 ~1722 (2nd to Polaris Alpha) · 64.78% user preference in silent rollout.

Independent: LMArena's community-run Text leaderboard independently confirmed Grok 4.1 Thinking's #1 placement at launch and its later displacement by Gemini 3 Pro (~1501). Third-party analyses (e.g. TheDayAfterAI) note the standout scores are human/model-judged preference tests, not traditional academic tasks — independent academic benchmark coverage is thin.