Elon Musk Said Grok 4 Was the “Smartest AI in the World,” But Its Leaderboard Scores Just Came Out and They Tell a Different Story

Elon Musk was proud of what he was saying are the amazing capabilities of the new Grok 4 Chatbot from Xai.

“Grok 4 is more intelligent than almost all graduate students in all disciplines, simultaneously,” Musk boasted, adding that GROK 4 was “the smartest artificial intelligence in the world.”

Is it really? Intelligence was difficult even before returning before it reached artificial intelligence, but some tests can provide some evidence.

One of the prominent platforms to do this is the Lamerna player who was developed at the University of California at the University of California, Berkeley, who gets sexual intercourse on artificial intelligence models by making users register their responses in categories ranging from creative writing and coding to mathematics and vision.

In its latest results, Grok 4 ranked third in general and the generation of the text. Don’t make mistakes, this is impressive – but it still fails to advance the advanced models of Google and Openai. (Specifically, the thinking models were placed in Gemini 2.5 from Google First and Openai’s O3 and 4O associated with the second, with GPT-4.5 with GROK 4 for the third.)

While Grok is clear that a frightening competitor in the arenas of racism and anti -Semitism, in other words, even its most recent version is no less than that it is “the smartest artificial intelligence in the world.” (This is not completely surprising; Musk has a long history of escaping in his career, political activities, and even his hobbies.)

Perhaps Grok’s only savings is the suggestion, according to expert criticism, that ARNA from Berkeley may be more than feelings than Scientific.

According to a recent study, conducted by a union of artificial intelligence researchers and his leadership of the COHERE machine, it is claimed that the leading painting has a set of “systematic issues that led to a distorted stadium.” Among the dangerous allegations raised by the researchers is the claim that the scene is conducting “unannounced special tests” before publicly released – and that the classifications can be withdrawn as desired.

Shortly after the release of the paper, it was revealed that the Llama 4 Meta version used by the leading painting was not the same that was publicly released-a trick of taste and switch on the side of Meta to charm human voters behind the arena.

Despite the release of an apology and threw Meta under the bus for its prominent attempts to manipulate the game, it was still a really bad look at Chatbot Arena. What does this mean for Grok, though? We will have to ask the smartest artificial intelligence in the world.

More about GROK: The Pentagon pumps 200 million dollars in AI Elon Musk, who had just the collapse of the Nazi