Elon Musk’s artificial intelligence startup, xAI, has unveiled its latest creation, Grok-3, a chatbot model that the entrepreneur claims is ‘scary smart’ and boasts significant improvements in preventing ‘AI hallucinations.’  During a live-streamed presentation alongside xAI engineers, Elon declared Grok-3 to be “the smartest AI on Earth.”

According to xAI, across a range of math, science, and coding benchmarks, Grok-3 outperforms Google’s Gemini, DeepSeek’s V3, Anthropic’s Claude, and even OpenAI’s GPT-4.

Grok-3 will initially be available to Premium+ paid X (formerly Twitter) subscribers before being rolled out to other users.  The upgraded chatbot enters a competitive AI landscape, with various companies and nations vying to create more sophisticated and cost-effective AI products.

Elon revealed that Grok-3 possesses “more than 10 times” the compute power of its predecessor and completed pre-training in early January. “Grok is to understand the universe,” he stated at the launch, emphasising the AI’s truth-seeking nature.

Grok-3’s advanced reasoning capabilities have allowed it to surpass the performance of any other released AI, according to Elon.  “Grok 3 has very powerful reasoning capabilities, so in the tests that we’ve done thus far, Grok 3 is outperforming anything that’s been released, that we’re aware of, so that’s a good sign,” he said in a video call with the World Governments Summit in Dubai. 

The chatbot is trained on synthetic data and employs self-correction mechanisms to avoid the ‘hallucinations’ that can plague other AI systems, preventing it from presenting false information as fact, according to sources.

Grok-3’s release pits Elon against his former collaborator and now rival, Sam Altman, and his OpenAI chatbot, ChatGPT.  Elon and Sam were among the founders of OpenAI in 2015, with the X owner initially providing USD 45 million in funding.  He departed in 2018, and in 2022, OpenAI’s release of ChatGPT became a global sensation, leaving Elon on the sidelines.