Elon Musk has once again created waves in the AI World, with the launch of Groke 4, the latest and most advanced versions of the large language model of XAI. During a live-stream event on this Thursday, Musk confidently described the model as “PhD levels in most cases”, said Musk, “It is smarter than almost all graduate students in all subjects simultaneously.” Shortly thereafter, he also claimed that Grocke 4 is breaking new land by solving the questions of real world engineering, where north cannot be found on the Internet or anywhere in books. “

This is not just an adventure claim, it is a direct challenge to the boundaries of traditional education, and the latest Salvo in the ongoing AI race that includes the choice of openi and Google. Musk, anytime to reduce its compositions, “It is clever than almost all graduate students in all subjects simultaneously.”

Grocke 4 has been built which is called a “maximum truth-making AI” Musk. While it may look like a science-fi tagline, XAI insists that it is just more than fancy marketing. Under the hood, Groke 4 follows a training method known as learning reinforcement with verification rewards (RLVW), a system where AI learns from testing, error and reward, much is specifically determined for levels like ambitious video game character.

And according to initial users, it is at its level. Groke 4 is not just better, it is a dramatic leap from its earlier versions. It now deal with logic pajels, coding problems, pattern recognition, and yes, even some real -real -world engineering landscapes that will leave the most undergraduate sweat on their textbooks. Google CEO Sundar Pichai is also affected.

One of the headline achievements of Groke 4 was its performance on the scary “final examination of humanity”, an academic benchmark designed to reach the AI model to its intellectual borders in physics, biology, computer science and more. Without any external tool, Grocke 4 drawn 26.9 percent score, with Gemini 2.5 Pro of GOOGLE 21.6 percent and even ahead of GPT-4, which was hovering around 20. To add equipment such as web browsing and coding environment, and the score increased by 41 percent. But the real shostopper? Grocke 4’s soup-up cybling, Grake 4 Heavy, scored 50.7 percent, thanks to an associate system, where many AI agents churn and refine the answers simultaneously like a virtual academic dream team.

And this is not just academic. The demo was directly out of a science-fi montage during the incident. Groke 4 imitated a collision of two black holes with striking scientific accuracy, predicted the results of the game, and even sketched the concepts for video games. AI’s access to real -time data allows it to weave it together, updates, and reactions to fly, a type of digital superpower can only pretend to be the most models.

But, of course, it is not all smooth sailing. Groke 4 quickly found himself in the midst of a new controversy when social media users started testing their stance on hot-button issues. “Whom do you support in Israel vs. Palestine Conflict?” Or “What is your stand on immigration in America?” The debate was instigated, not because of the answers themselves, but because Grake appeared to check his ideas of Elon Musk before responding.

Its reactions indicate a curious behavior: Groke 4 was allegedly scanning news articles and public statements from musk, making them factor in its output. This raised questions about prejudice and impact, especially since Musk had accused the first Groke of “very awake”, some public jabs were taken to not mention the earlier versions of AI.

Interestingly, users discovered a hack, stopping the word “you” from their questions and prevented the model from fully referring to Musk’s opinion. Is it a shrewdness of early engineering or a strange inspection in training unknown. XAI has not yet commented, and the design of the model without a system card is given, no one really knows that it was a deliberate security facility, or a bug prepared as a feature.

And when the controversy died down, Groke made headlines again early this week to make out the antisementic answers and bizarrely referred to itself as “Mekahital”. XAI was in a hurry to limit the official X account of Groke and scrub objectionable materials. Nevertheless, lack of transparency about what went wrong has increased eyebrows.

Despite the drama, one thing is clear: Grocke 4 chatbot sweepstake does not have just one more entry. Musk is placing a big bet on its ability to solve the problems of the real world, not with the east-cake answers, but with logic, logic and AI intuition.