Grok 4.5 enters private beta, Elon Musk claims performance on par with Claude Opus

Elon Musk says xAI's upcoming Grok 4.5 has entered private beta testing at Tesla and SpaceX, with internal evaluations suggesting performance comparable to Anthropic's Claude Opus as the company accelerates AI development and infrastructure upgrades.

xAI is preparing to launch the next version of its artificial intelligence chatbot, Grok 4.5, with founder Elon Musk claiming the model could rival Anthropic's Claude Opus. The latest version has entered private beta testing at Tesla and SpaceX, marking another step in xAI's efforts to strengthen its position in the increasingly competitive AI market.

In a series of posts on X, Musk said Grok 4.5 is powered by xAI's new V9 foundation model, which features around 1.5 trillion parameters. He added that the model has also received supplementary training using data from the AI coding platform Cursor. Based on internal evaluations, Musk said Grok 4.5 performs at a level comparable to Anthropic's Claude Opus and may even surpass it in certain areas. However, he clarified that he was not suggesting the model would outperform every competing AI system, describing it instead as "a solid workhorse in the same league as Opus."

xAI speeds up AI development

Musk said reinforcement learning continues to improve Grok's capabilities while the company's development process has become more efficient. He added that SpaceX plans to release entirely new AI foundation models trained from scratch every month for the rest of the year, signalling a significantly faster development cycle than before.

According to Musk, Grok 4.5 represents a major improvement over the previous V8 foundation model, which contains around 500 billion parameters and has several architectural limitations. He also credited engineers from SpaceX's Starlink and Starship programmes for contributing to xAI's AI efforts, saying several senior aerospace engineers have shifted much of their focus towards artificial intelligence development.

Focus on faster infrastructure

Besides improving model performance, xAI is also redesigning the software infrastructure behind Grok. Musk said the company is rewriting key training and inference systems in C and C++ to simplify the software stack and improve efficiency. He added that Grok is being optimised for Nvidia's GB300 AI platform to deliver better training performance and faster inference.

The latest roadmap indicates that xAI is combining larger AI models with software optimisation and a faster release schedule to compete with leading AI developers. Since Grok debuted in 2023, it has become central to Musk's AI strategy and has been integrated into the X platform while being tested across his wider group of companies.

While Musk's performance claims have not been independently verified, the announcement suggests xAI is aiming to accelerate innovation as competition among AI companies intensifies.