SpaceXAI introduces Grok 4.5, marking its inaugural model featuring Cursor.

      SpaceXAI has unveiled Grok 4.5, its most advanced model to date. This launch marks the company's first release since going public and acquiring the AI coding startup Cursor. This model represents the collaborative effort of both firms aimed at coding and agentic tasks rather than casual conversation.

      “It is an Opus-class model, but faster, more token-efficient, and cost-effective,” Musk stated on X. The model name references Anthropic’s leading Opus family. A chart accompanying the announcement indicates that Grok 4.5 surpasses Opus 4.8 in several benchmarks, as first reported by Axios.

      Designed for programmers and the financial sector, Grok 4.5 was developed in conjunction with Cursor, which SpaceX has agreed to acquire for a valuation of $60 billion. According to the company’s blog, this model is intended to “manage complex, long-duration tasks,” such as software engineering. Unlike Cursor’s previous models, it also addresses legal and financial tasks and incorporates cybersecurity capabilities, as reported by Bloomberg.

      The push into finance is intentional. Musk mentioned earlier this year that his AI division, previously known as xAI before merging with SpaceX, had fallen behind in coding. The company has since restructured its team and sought Wall Street clients for its Grok chatbot. Grok 4.5 is the most significant indicator of this shift towards serving paying business customers.

      The pricing strategy is attractive. Grok 4.5 is offered at $2 per million input tokens and $6 per million output tokens. In comparison, Anthropic’s Opus 4.8 charges $5 and $25, while OpenAI’s GPT-5.6 Luna is priced at $1 and $6. This pricing undercuts the more expensive competitors as companies closely monitor their token expenditures.

      The company does acknowledge some limitations. It states that Grok 4.5 outperforms certain models from OpenAI and Anthropic in speed, cost, and performance, but does not surpass their largest and latest offerings. Musk anticipates closing that gap soon. The model is now live in Grok Build, available in Cursor across all plans, and accessible via the SpaceXAI console, with a broader public release expected on Thursday. It is currently not available in the EU.

      The timing of this release is significant as it coincides with OpenAI's extensive rollout of GPT-5.6 and a new suite of voice models on Thursday, following requests from the Trump administration to stagger the launch. There is heightened governmental scrutiny, as regulators monitor new models for cybersecurity risks. Cursor has stated that it has implemented measures to “detect and block bad actors” while ensuring legitimate security research is maintained.

      There is an interesting aspect regarding the development of Grok 4.5. SpaceXAI trained the model using the same computing resources it leases to competitors Anthropic and Google. As its models require more resources, SpaceXAI will face a choice: to allocate those resources for its own use or rent them out for revenue. For the moment, Musk believes that a more affordable and coder-friendly Grok can attract business, despite its current gap in raw performance compared to leading models.