xAI has officially unveiled Grok 4.1, the newest version of its large-language model and the most substantial update since the company introduced the Grok 4 family earlier this year. The model is now available to all users on Grok.com, the mobile apps and across X’s interface, deploying immediately in Auto mode with an option to manually select the new version.
“We are excited to introduce Grok 4.1, which brings significant improvements to the real-world usability of Grok. Our 4.1 model is exceptionally capable in creative, emotional, and collaborative interactions. It is more perceptive to nuanced intent, compelling to speak with, and coherent in personality, while fully retaining the razor-sharp intelligence and reliability of its predecessors. To achieve this, we used the same large scale reinforcement learning infrastructure that powered Grok 4 and applied it to optimize the style, personality, helpfulness, and alignment of the model. In order to optimize these non-verifiable reward signals, we developed new methods that let us use frontier agentic reasoning models as reward models to autonomously evaluate and iterate on responses at scale,” the AI firm announced in an official statement.
Prior to the official reveal, Grok 4.1 underwent a two-week silent rollout from November 1 to 14, during which xAI diverted a portion of live user traffic to the model without announcing the change. According to the company, Grok 4.1 achieved a 64.8% win rate in blind, pair-wise comparisons against the previous production release, indicating stronger perceived quality in real-world use. In comparison, GPT-5.1 remains stronger in complex coding and enterprise-grade tool use.
The model also posted significant gains on public leaderboards. On the LMArena Text Leaderboard, the model’s reasoning variant, internally code-named “quasarflux,” recorded an Elo score of 1483, placing it at the top of the ranking and ahead of systems released by competing labs. Its non-reasoning variant, known as “tensor,” scored 1465, placing second overall and surpassing many models that include full reasoning capabilities. Other benchmarks show similarly large jumps. Grok 4.1 scored 1586 on EQ-Bench3, a test of emotional intelligence, and 1722 on Creative Writing v3, both record highs for the company and substantial improvements over Grok 4’s results.
Beyond benchmark strength, xAI is positioning Grok 4.1 as a more capable everyday assistant. The company said the system’s emotional intelligence improvements are not cosmetic but result from a deeper modeling shift. Using agentic reasoning models as evaluators, Grok 4.1 receives reward signals tied to empathy, contextual reading of intent, and tonal appropriateness. The model is also designed to deliver a more consistent personality and writing style. xAI said users should see fewer abrupt shifts in tone between responses, improved stability in long conversations, and a stronger ability to follow user-defined voice or character constraints.
Hallucination rates have been reduced as well, according to internal assessments. xAI estimates that Grok 4.1 is three times less likely to provide fabricated information compared with the prior generation. The reduction stems partly from new error-detection mechanisms inside the reinforcement-learning stack, as well as expanded real-time evaluation during training. xAI is following the same distribution strategy it used for earlier Grok releases, offering the upgraded model to all users — including free accounts — without paywalls or usage tiers that gate the higher-performing version. The company said both Grok 4.1 and the enhanced Grok 4.1 Thinking variant are available at launch, though paid users continue to receive higher usage limits. Grok 4.1 is available immediately across all platforms.
The Tech Portal is published by Blue Box Media Private Limited. Our investors have no influence over our reporting. Read our full Ownership and Funding Disclosure →
