The world's most powerful AI model: Musk launches Grok 4, which can handle academic problems at the PhD level. During a live broadcast held today, Elon Musk announced the launch of Grok 4, claiming to be the world's strongest AI model, whose performance in handling academic problems has reached the PhD level. The entire conference lasted 53 minutes, and Musk demonstrated many new features such as the multi-modal capabilities and faster reasoning capabilities of the Grok 4 model. Judging from the official running score data, the Grok 4 model has very impressive results. In terms of real-time speed, reasoning capabilities and advanced vision, it will compete with top models from OpenAI and Anthropic. The Grok 4 model is capable of deep reasoning and has been trained on xAI s Colossus supercomputer, promising greater logical reasoning and text generation capabilities. Grok 4 also features natural human speech characteristics, real-time network access, and the intelligence to understand Internet culture, including high-precision recognition of memes, slang, and humor. According to the running score results published by Artificial Analysis, the intelligence index of Grok 4 is 73. For comparison, OpenAI's o3 model is 70, Google Gemini 2.5 Pro model is 70, Deepseek R1-0528 model is 68 points, and the Anthropic Claude 4 Opus model is 64 points. Grok 4 is an inference model that will be provided through the xAI API. In terms of pricing, the input price is US 3 per 1 million tokens (IT House Note: the current exchange rate is approximately 21.5 yuan), and the output price is US 15 per 1 million tokens (the current exchange rate is approximately 108 yuan)