Elon Musk’s xAI Unveils Grok 4.1, Promises AI Is 3x Less Likely to Fabricate Information

Elon Musk-led AI startup xAI has launched its latest AI model, Grok 4.1, promising faster performance, improved reliability, and significantly reduced errors in generating false information. The model has shown strong results in benchmarks and is now available across multiple platforms.

xAI Launches Grok 4.1: What’s New?

xAI, the artificial intelligence company founded by Elon Musk, has unveiled Grok 4.1, the newest iteration of its AI language model. Musk shared the update on his social media platform X, emphasizing noticeable improvements in speed and output quality. The latest model focuses on minimizing “hallucinations,” a common issue in AI where false information is presented as fact.

Musk stated on X: “Grok 4.1 just released. You should notice a significant increase in speed and quality.” According to xAI, the model retains the intelligence and reliability of its predecessors while enhancing creative, emotional, and collaborative interactions.

Grok 4.1 Reduces Hallucinations by 3x

A key feature of Grok 4.1 is its ability to reduce factual errors. xAI reports that the new model is three times less likely to produce false information compared to Grok 4 Fast. Post-training adjustments focused on improving accuracy for information-seeking queries, which have traditionally been prone to hallucinations in large language models (LLMs).

Real-world testing and the FActScore benchmark, which evaluates AI on 500 biography-related questions, revealed a marked improvement. While the previous Grok 4 Fast model had a hallucination rate of 12%, Grok 4.1 reduced it to just 4%. Similarly, on the FActScore test, the model scored 2.97% error compared to 9.89% in its predecessor, demonstrating a significant enhancement in reliability.

Grok 4.1 Tops AI Benchmarks

xAI also tested Grok 4.1 on LMArena, a widely recognized benchmark for LLMs. In the Text Arena category, Grok 4.1 in quasarflux mode achieved the highest overall Elo score of 1483, outperforming all non-xAI models by 31 points. Even in its non-reasoning tensor mode, the model ranked second, scoring higher than other models’ full-reasoning setups.

These results indicate that Grok 4.1 not only improves factual accuracy but also maintains state-of-the-art performance in reasoning, problem-solving, and general AI tasks.

Gradual Rollout and User Feedback

Grok 4.1 underwent a two-week soft launch from November 1 to 14, 2025, allowing xAI to introduce the model to a small group of users while conducting live blind pairwise evaluations. During this period, the AI’s responses were compared against the previous version, and Grok 4.1 achieved a win rate of 64.78%, highlighting a strong user preference for the upgraded model.

This staged rollout allowed xAI to fine-tune performance in real-world conditions before the full public release.

How to Access Grok 4.1

Grok 4.1 is now available to all users across platforms, including:

• Web: Accessible via grok.com

• Social Media: Integrated into X

• Mobile Apps: Available on both iOS and Android

Users can select Grok 4.1 in the mode picker within the app or use Auto mode for an optimized experience.

Key Features of Grok 4.1

• Reduced Hallucinations: 3x lower likelihood of generating false information

• Enhanced Speed & Quality: Faster responses with improved output

• Creative & Collaborative Interactions: Supports emotional and context-aware dialogue

• High Benchmark Scores: Top-ranked in LMArena tests for large language models

• Cross-Platform Access: Available on web, mobile apps, and X

Conclusion

With Grok 4.1, xAI aims to set a new standard in AI reliability, reducing misinformation while improving user interaction. The launch strengthens xAI’s position in the AI market and provides users with a faster, more accurate, and versatile AI experience.