According to foreign media reports, the AI model Grok 4, developed by Elon Musk's AI startup xAI, has reportedly achieved outstanding results in leaked test data. In the "Humanities Last Exam" (HLE), Grok 4 scored 35%—a performance that improved to 45% with the use of reasoning techniques. These results significantly surpass the previous highest score of 20% achieved by ChatGPT o3.
If the leaked data proves accurate, this suggests that Grok 4 may outperform other leading AI models, including Gemini 2.5 Pro, ChatGPT o3 Pro, and Claude 4 Opus.
The report also highlighted that although Musk had previously hinted at a release date for Grok 4 shortly after July 4, the model has yet to make an appearance. However, with competitors such as OpenAI, Google, and Anthropic recently announcing new AI products, xAI is under increasing pressure. Analysts predict that Grok 4 is highly likely to go live next week.
Related News:
Elon Musk launches 'America Party,' Trump calls it 'ridiculous'
Tesla struggles with slumping sales as Musk's longtime aide departs
Comment