Elon Musk's Grok 3 AI Model Claims to Outperform GPT-4 in Key Benchmarks

The newly launched Grok 3 introduces advanced reasoning modes and real-time data capabilities but faces scrutiny over its true reasoning abilities.

Grok 3, developed by Elon Musk's xAI, has been introduced as a competitor to OpenAI's GPT-4, boasting improved reasoning and real-time data analysis features.
The model includes two unique reasoning modes: 'Think,' which displays its thought process, and 'Big Brain,' designed for complex computations.
Early tests suggest Grok 3 outperforms GPT-4 in some areas, such as real-time data analysis, but its creative and reasoning abilities remain comparable or slightly better in specific cases.
Spanish researchers have identified flaws in AI benchmarks, revealing that models like Grok 3 often rely on memorization rather than genuine reasoning, with performance dropping significantly under modified testing conditions.
Grok 3 is currently in beta and available for free temporarily, with a premium subscription required for future access to advanced features.