Elon Musk Introduces Grok 4

Elon Musk’s company xAI has officially introduced the fourth version of its artificial intelligence system — Grok 4. The new model is capable of solving PhD-level problems and comes in two versions: the standard Base model and Grok 4 Heavy, an enhanced version that operates using a team-based architecture with multiple bots working collaboratively.

Grok 4 scored 50.7% on the well-known Humanity’s Last Exam, outperforming Google Gemini 2.5 Pro (22%) and OpenAI’s GPT-3.5 (21%). The Heavy version, equipped with additional tools, achieved 44.4%, significantly higher than Gemini’s 26.9%.

The key innovation in Grok 4 is its collaborative AI architecture, where multiple AI assistants work on a task simultaneously, exchange results, and select the best outcome.

The launch of Grok 4 comes shortly after reputational concerns linked to antisemitic comments made by the automated Grok account on X (formerly Twitter). Despite its technical achievements, such incidents could hinder Grok 4’s adoption among corporate clients, for whom reliability and predictability are critical.