Elon Musk’s AI Company, XAI, released its latest flagship AI model, Grocke 4 late on Wednesday, and unveiled a new $ 300-maheen’s AI subscription plan, Supergroau Heavy.
Groke is an answer to Xai for Openi’s chat and Google’s Gemini, and analyze images and answer questions. In recent months, Groke Musk’s social network, X has been more deeply integrated into the X, which was recently acquired by XAI. However, it has also placed the misconduct of groukes at the center for millions of users.
Expectations for Grok 4 are high. The latest AI model from XAI will be piled up against Openai’s upcoming AI model, GPT-5, which is expected to be launched later in this summer.
Elon Musk said during a livestream on Wednesday night, “In relation to academic questions, Groke 4 is better than PhD level in every subject, no exception.” “Many times, it may lack general knowledge, and it has not yet invented new techniques or has discovered new physics, but it is only a matter of time.”
The launch of Groke 4 comes in between a week for Elon Musk’s companies. Earlier on Wednesday, Linda Yacarino stepped into his role as CEO of X almost two years with the company. The X has not announced its successor yet.
Yachcarino’s departure comes a few days after the official, automatic X account of the Groke, criticizing the “Jewish authorities” of Hollywood and praising Hitler and responded to users with antisementary comments. XAI had to briefly limit Groke’s account and remove aggressive positions. In response to the incident, XAI has removed the section recently added from the public system prompt of Groke, to follow a list of instructions for the AI Chatbot, which said that it is not reluctant to make “politically wrong” claims.
Instead of focusing on the performance and capabilities of Grocke 4, the leaders of Musk and XAI avoided discussing the event to a great extent.
XAI launched two models on Wednesday: Grocke 4 and Groke 4 have a “multi-agent version” of the company in heavy-cloth that provides an increase in performance. Musk claimed that the Groke 4 Heavy sponsored several agents to work on a problem together, and then they all compare their work to “like a study group”, which is to find the best answer.
XAI claims that Groke 4 reflects the performance of the frontier level on several benchmarks, including the final examination of humanity- a challenging test that measures AI’s ability to answer thousands of crowd questions on topics such as mathematics, humanities and natural sciences. According to XAI, Groke 4 scored 25.4%in the final examination of humanity without “tools”, which was performing better than Google’s Gemini 2.5 Pro, which scored 21.6%and O3 (high) of OpenaiI, which scored 21%.
XAI claims that Grake was able to get a score of 44.4%, with 4 heavy, “tools”, with the tool better than Gemini 2.5 Pro, which scored 26.9%.
Non -profit arc award states that Grake receives a new state -of -the -art score Its ARC-AGI-2 test-one and hard benchmark that have problems like puzzle, where AI has to identify the visual pattern -16.2%scoring. It is almost twice the next best commercial AI model, Claude Opus 4.
Along with Groke 4 and Groke 4 Heavy, XAI launched its most expensive AI subscription plan, which is a $ 300-month subscription, called Supergroau Heavy. Customers of the scheme will get an initial preview to make 4 heavy groukes, as well as reach new facilities. The scheme is similar to the ultra-pramium tires offered by Openai, Google and Anthropic, but Xai now offers the most expensive membership among major AI providers.
Some new products can reach the XAI schemes to launch supergroke heavy customers in the coming months. The company said on Wednesday that an AI coding model is coming in August, a multi-model agent in September and a video generation model in October.
The XAI model is releasing Grocke 4 through its API in an attempt to achieve developers to create applications. The company notes that XAI’s venture area is only two months old, however, it is planning to work with hypersscalers to provide groukes through its cloud platform.
Despite Groke’s Frontier-level performance on the benchmark, it can prove to be difficult for XAI to pursue its recent accidents as it tries to pitch the grouke for businesses as a real contender for chat, cloud and Gemini. Whether business groukes, flaws and all are ready to adopt, it remains to be seen.
