xAI said Grok 4.1 is better at understanding people’s feelings, being creative, and having smoother interactions.
- Elon Musk called the Grok 4.1 version a ‘major’ upgrade.
- The update preserves Grok 4’s strong reasoning while making the model more emotionally aware and conversational.
- The non-reasoning version of Grok 4.1 reduced its hallucination rate to about 4.22%.
Elon Musk called the latest version of Grok, the AI model developed by his company, xAI, a huge update from its previous rendition.
In a post on X, Musk replied to a post listing Grok 4.1’s improvements.
What’s New In Grok 4.1
The company said Grok 4.1 is better at understanding people’s feelings, being creative, and having smoother interactions. It’s also more reliable and quicker, thanks to improved training techniques.
According to xAI, the update preserves Grok 4’s strong reasoning while making the model more emotionally aware and conversational.
xAI built Grok 4.1 using the same large-scale reinforcement learning framework as Grok 4, but added a new reward system. This system uses reasoning “agent” models to rate Grok’s responses.
Grok 4.1 is available on grok.com, the X app, and iOS and Android devices.
Performance Gains
During a two-week ‘silent rollout’ from November 1 to November 14, xAI ran blind comparisons on live traffic. Grok 4.1 won 64.78% of the time when compared to the previous version.
In creative writing, it performed strongly on a benchmark that assesses style, humor, and narrative structure over 32 prompts. xAI also reported a significant drop in ‘hallucinations’, cases where the AI confidently states incorrect facts.
The non-reasoning version of Grok 4.1 reduced its hallucination rate to about 4.22%, compared to roughly 12% in the earlier Grok 4. xAI introduced Grokipedia in October, an AI-built encyclopedia created using Grok.
Grok 4.1 has been positioned as a competitor to other famous AI models, including OpenAI’s ChatGPT and Google’s Gemini.
For updates and corrections, email newsroom[at]stocktwits[dot]com.<