SpaceX has unveiled a new AI model. This is its first joint development with Cursor

On July 8, SpaceXAI unveiled a new artificial intelligence model, Grok 4.5, developed in collaboration with the startup Cursor, which it announced it would acquire following its IPO. Elon Musk’s company hopes to strengthen its position in the corporate AI market by offering a tool for financial and legal tasks, as well as programming.

Grok 4.5 became the first joint AI model developed by SpaceXAI—which was known as xAI prior to its merger with SpaceX—and the startup that created Cursor, a popular AI code-writing assistant. It, too, is set to become part of Musk’s company soon—the deal was agreed upon in June.

The new model significantly expands the capabilities of previous Cursor solutions, SpaceX said. It is designed to perform complex and time-consuming tasks that require the analysis of large amounts of information and multi-stage processing. In addition to generating program code, Grok 4.5 is adapted to work with financial and legal documents and has enhanced cybersecurity capabilities.

The model incorporates mechanisms for detecting and blocking attackers. According to Cursor, these mechanisms preserve opportunities for legitimate cybersecurity research—including vulnerability discovery and remediation—while restricting use cases that could be dangerous, Bloomberg reports.

This is another step by SpaceXAI toward developing its line of enterprise AI products, the agency notes. Musk had previously acknowledged that his developers were lagging behind competitors in this area, after which he began a major restructuring of the team.

According to Bloomberg, SpaceX is seeking to expand its presence in the financial sector and attract clients from Wall Street, with the aim of increasing revenue from corporate AI services.

The release of Grok 4.5 has been accompanied by increased scrutiny from authorities regarding new AI models, particularly their capabilities in the field of cybersecurity. The U.S. government temporarily restricted foreign users’ access to Anthropic’s most powerful models, Fable and Mythos, and also asked OpenAI to delay the release of its latest GPT-5.6. Both AI giants have already announced that they have received approval to launch their products.

This article was AI-translated and verified by a human editor