Grok 4.1
2025-11-17 · Publisher : xAI · Domain : Artificial Intelligence
xAI rolls out Grok 4.1 to all users of grok.com, X and the mobile apps, with a more consistent personality, more emotional intelligence and fewer hallucinations.
What changes
- Improved style, personality and alignment through reinforcement learning, with agentic reasoning models used as reward models.
- Silent rollout from November 1 to 14, 2025: Grok 4.1 is preferred in 64.78% of blind comparisons against the previous production model.
- Reduced hallucination rate on information-seeking queries (evaluation on real traffic and FActScore).
- Two modes: Grok 4.1 (without thinking) and Grok 4.1 Thinking.
### Specifications - Availability: grok.com, X, iOS, Android; select “Grok 4.1” in the model picker or Auto mode. - Context, maximum output, knowledge cutoff and API identifier: not disclosed in the post (the API models arrive with Grok 4.1 Fast). ### Pricing Pricing not disclosed at launch (use within the apps). ### Benchmarks Scores published by the vendor. | Benchmark | Result | |---|---| | LMArena Text Arena, Grok 4.1 Thinking | 1st, 1483 Elo | | LMArena Text Arena, Grok 4.1 (without thinking) | 2nd, 1465 Elo | | Grok 4 (comparison rank) | 33rd | EQ-Bench3 and Creative Writing v3 scores are cited but only in charts. ### Key takeaways An update focused on conversational quality and factual reliability rather than pure reasoning.
Links
Tags : LLM, xAI, Grok, Grok 4.1