Summary

  • Elon Musk’s AI company, xAI, has released a new AI model, Grok 3, which it claims outperforms leading competitors according to key technical benchmarks.
  • The model has better scores in mathematics, scientific reasoning and coding tasks than OpenAI’s GPT-4o, Google’s Gemini and DeepSeek’s V3 model.
  • One of the key innovations in Grok 3 is the DeepSearch feature, which combines web searching with reasoning capabilities to analyse information from multiple sources.
  • Grok 3 has a ‘Think’ function, which shows the model’s reasoning process, and a ‘Big Brain’ mode to allocate more computing power to difficult tasks.
  • However, early tests show that it sometimes fabricates citations and struggles with certain types of humour and ethical reasoning tasks.

By Michael Nuñez

Original Article