Elon Musk’s artificial intelligence company, xAI, has officially launched its latest AI model, Grok 3, claiming it outperforms OpenAI and China’s DeepSeek based on early testing in math, science, and coding. The announcement was made during a live demonstration streamed on Musk’s social media platform, X, where he emphasized that Grok 3 is significantly more advanced than its predecessor.
Musk, speaking at the event, described Grok 3 as an “order of magnitude” more capable than Grok 2, boasting powerful reasoning abilities. The model, he said, has been trained on a vast amount of synthetic data, allowing it to recognize and correct its own mistakes, leading to greater logical consistency. He also teased that this could be the last time any AI model surpasses Grok.
Alongside Grok 3, xAI introduced “Deep Search,” a next-generation AI-powered search engine. The new model will be made available to premium X subscribers later in the day, with additional access through a separate subscription for the web and app versions.
At last week’s World Governments Summit in Dubai, Musk had already hyped up Grok 3, calling it “scary smart” and asserting that it had outperformed all existing AI models in xAI’s internal tests. The xAI team also cited independent evaluations, stating that an early iteration of Grok 3 received higher ratings than its competitors on Chatbot Arena, a crowdsourced platform that pits AI models against each other in blind tests.
During the product demo, Musk reassured users that improvements would continue at a rapid pace. He described Grok 3’s launch as a “beta phase,” warning of minor imperfections initially but promising frequent updates, with voice assistance for the model set to roll out later.
Read Also: Elon Musk’s shocking $97B bid to buy OpenAI rejected by Sam Altman
Musk’s entry into the generative AI market with xAI in 2023 has fueled intense competition, particularly with OpenAI’s ChatGPT. OpenAI launched its most advanced model, the o1, last September, enhancing its reasoning capabilities to tackle complex coding, math, and science problems.
Musk, who co-founded OpenAI in 2015 alongside Sam Altman, has been in a heated feud with the company’s leadership in recent years. He recently led an investor group that offered $97.4 billion to acquire OpenAI’s nonprofit parent organization—a bid that was swiftly rejected.
Adding to the AI arms race, China’s DeepSeek stunned the industry last month by claiming that one of its open-source models could rival OpenAI’s o1 while using a more cost-effective and energy-efficient training process. This breakthrough came despite U.S. restrictions preventing Nvidia from selling its most advanced AI chips to China.
xAI has been ramping up its own computing power, revealing that it has doubled its GPU cluster to train Grok 3. Last year, the company had already deployed a “Colossus supercomputer,” featuring 100,000 Nvidia GPUs.
While DeepSeek’s rise has been hailed by some experts as a game-changer for AI, others remain skeptical about its long-term impact. Meanwhile, xAI’s latest move signals that Musk is determined to push the boundaries of AI development, setting the stage for an even fiercer battle in the race for artificial intelligence supremacy.
