xAI, founded by Elon Musk, has unveiled a new version of its artificial intelligence called Grok 4. This AI is marketed as the most powerful in the world and is capable of solving complex engineering problems.
Description of Grok 4 and Its Capabilities
Grok 4 has been described as a model that nearly never makes mistakes when solving math and physics problems, apart from intentionally designed tricks. "Grok 4 is at a level where it almost never gets math and physics questions on exams wrong," Musk stated. He also noted that this AI can identify errors or ambiguities in questions and either correct them or provide answers for all possible interpretations.
Testing Results and Comparisons with Competitors
Grok 4 delivered high performance in various tests, including Humanity's Last Exam, where it scored 25.4%, surpassing Google’s Gemini 2.5 Pro (21.6%) and OpenAI O3 (21%). In the ARC-AGI-2 benchmark, Grok achieved an average score of 16.2%. This test involves puzzle tasks where AI must recognize visual patterns, showcasing its versatility.
SuperGrok Heavy Subscription and Its Benefits
xAI also introduced the new SuperGrok Heavy subscription plan priced at $300 per month, granting access to the high-performance version of Grok and early trials of xAI products. Subscribers will receive features such as advanced programming tools, prioritized technical support, and increased usage limits. Additional functionalities like DeepSearch, Grok Studio, and Big Brain will also be available.
The launch of Grok 4 by xAI highlights Elon Musk's ambition to create powerful artificial intelligence tools capable of solving complex problems and competing with existing models. However, the new release also faces some challenges and criticism.