Musk’s xAI launches Grok 3 AI model, claims superior performance

USA – Elon Musk’s xAI has introduced its latest artificial intelligence model, Grok 3, claiming it outperforms rivals like OpenAI and China’s DeepSeek in early testing.

The tests, which included math, science, and coding tasks, showed that Grok 3 exceeded expectations.

“We’re thrilled to present Grok 3, which we believe is much more capable than Grok 2 in a very short time,” Musk shared during the model’s demonstration, which was streamed on his social media platform, X. The team also unveiled a new product called “Deep Search,” a next-generation search engine.

Starting Tuesday, Grok 3 will be available to premium X subscribers in the U.S. and can also be accessed through a separate subscription for the web and app versions, according to the xAI team.

Musk recently described Grok 3 as “scary smart,” with powerful reasoning skills. At the World Government Summit in Dubai, he stated that the AI model outperformed all others in xAI’s internal tests.

Musk added, “This might be the last time an AI is better than Grok,” emphasizing that it was trained using vast amounts of synthetic data and had the ability to learn from its mistakes to ensure logical consistency.

In early tests on Chatbot Arena, a crowdsourced platform that compares AI models, Grok 3 received higher ratings than existing competitors.

Musk highlighted that the model was still in its beta phase, meaning it might have some imperfections, but improvements would come rapidly. He also teased the release of voice assistance features at a later stage.

The competition in the AI space is fierce. Musk, who co-founded OpenAI in 2015, entered the generative AI market in 2023 with xAI.

Last year, OpenAI launched its advanced o1 model, which also demonstrated strong reasoning and problem-solving abilities in areas like science and coding.

However, Musk and OpenAI’s leadership have had disagreements in recent years, especially after Musk attempted to acquire OpenAI for US $97.4 billion, an offer the company declined.

In addition to OpenAI, the Chinese startup DeepSeek recently released a technical paper showcasing its model’s ability to rival OpenAI’s o1, despite using a less energy-intensive process. DeepSeek’s success has sparked discussions on how AI can be trained with lower-cost technology.

xAI has been ramping up its efforts, with its “Colossus supercomputer” now using a cluster of 100,000 advanced Nvidia GPUs for AI training.

On Tuesday, the company revealed that it had doubled the size of its GPU cluster to train Grok 3. While DeepSeek has raised the level of competition in AI, some experts remain skeptical about its impact.