Elon Musk’s New Artificial Intelligence: The billionaire unveiled version 4 of his AI engine Grok, in what he calls the smartest engine in the world. This launch, which he says marks the biggest leap so far in the field of artificial intelligence, took place just hours after the company’s bot sparked a global controversy due to severe antisemitic statements, including praise for Hitler.
During a one-hour live launch event, Musk, together with engineers from xAI, presented the new capabilities of Grok 4. According to him, the engine passed the “final test of humanity,” which included thousands of questions written by users in fields such as mathematics, humanities, and exact sciences. Musk claimed that Grok 4’s results surpassed those of competing AI engines like Google’s Gemini 2.5 Pro and OpenAI’s top-tier ChatGPT version.
During the event, Musk showed two separate versions of Grok 4: A regular version and a more powerful “Heavy” version. The first will cost users $30 per month, while the latter is offered to paying subscribers at $300.
Musk did not hold back on self-praise: he said that Grok 4 is “smarter than almost all advanced degree students, in all fields, simultaneously.” According to him, if the engine were tested on the SAT exam, it would get a perfect score every time. He even added that Grok 4 answers academic questions at a level higher than a doctorate “in every field without exception.”
However, Musk admitted that the engine might sometimes suffer from a lack of common sense and noted that in some areas it is “still a child-genius who needs to be instilled with the right values.” Musk predicted that by the end of this year or at the latest next year, Grok will be able to “discover new technologies that will be truly useful,” and even estimated that the engine might “discover new physics” as soon as the coming year.
Although the event took place only hours after the previous Grok system posted messages expressing support for Hitler and presenting itself by default under the name “MechaHitler,” Musk did not address those statements directly. However, he did refer to the general issue of AI safety and said: “We need to make sure that AI will be good AI, good Grok.”
According to him, “The most important thing in AI safety is to strive for maximal truth-seeking. You can think of AI as a child-genius who will eventually surpass you, but you can instill good values in it, encourage it to be honest, respectful, and strive for good—just like a smart and powerful child should be.”
xAI, which quickly removed the antisemitic posts after the storm, issued a statement saying it had acted to remove hate from the platform even before the messages were published. The company also changed the internal guidelines that direct the Grok engine when formulating its responses. Among other things, instructions that required the engine not to fear expressing “non-politically correct opinions, as long as they are based [on facts],” were canceled.
Meanwhile, Grok 4 has already been launched to early subscribers on the X platform (formerly Twitter), and the internet is buzzing with the question of whether the new engine truly lives up to its promises or whether it will also need to go through rounds of improvements following serious errors or deviations from accepted norms.