How good is xAI's Grok 4?

Frontier models keep showing fairly considerable improvements. 🚀

Hey Everyone,

While Nvidia has hit officially a $4 Trillion market cap, Elon Musk’s xAI launched their new frontier model Grok 4 on July 9, 2025. A frontier model just 28 months after xAI was founded.

I’m not a huge fan of the utility of LLM benchmarks, but according to artificial analysis.ai Grok 4 is indeed the smartest frontier model. For at least a few weeks.

xAI has made stunning progress in the 500 days since they released their first model.

From what we know Grok 4 has 1.7 trillion parameters with a hybrid architecture that has specialized modules for different cognitive tasks. It’s powered by xAI's Colossus supercomputer with 200,000 Nvidia GPUs.

How to Access Grok 4

Grok 4 is now available through three main entry points: the X app, the xAI API, and the grok.com platform.

Evolution of Grok in a Nutshell

On Grok 4: “It's a reasoning model where you can't see the reasoning tokens or turn off reasoning mode.”

Simon Willison

- his first take.

Grok 4 describes the above image as: a "cute, bird-like creature (resembling a duck, chick, or stylized bird)".

• In this article we’ll go deeper into Grok 4’s frontier aspects related to benchmarks.

• We’ll talk about the potential of Grok 4 and try to describe Grok 4 Heavy.

• We’ll cover pros and cons, capabilities and limitations of Grok 4.