SpaceXAI has launched Grok 4.5, the first model developed post-rebranding from xAI, in collaboration with AI company Cursor. Grok 4.5 is designed specifically to excel at coding, agentic tasks, and knowledge work, according to the company. The model was trained on tens of thousands of NVIDIA GB300 GPUs utilizing extensive datasets in coding, science, engineering, and math.
SpaceXAI claims Grok 4.5 outperforms other leading models in real engineering tasks, showing proficiency at creating functional applications with minimal instructions. An example highlighted by SpaceXAI includes Grok 4.5 generating an interactive simulation of the solar system from a single prompt. The model is claimed to operate faster than “flash” models while delivering results at lower costs, priced at $2 per million input tokens and $6 per million output tokens.
For context, OpenAI’s GPT-5.6 variant Sol costs $5 per million input tokens and $30 per million output, while its Luna variant is priced at $1 per million input tokens and $6 per million output. Grok 4.5 is now the default model for SpaceXAI’s terminal-based AI coding agent, Grok Build, which can also handle tasks in Excel, PowerPoint, and Word.
Access to Grok 4.5 is available through the SpaceXAI console and included in all Cursor’s plans. In April, SpaceXAI and Cursor announced a partnership focused on AI development. This partnership may involve SpaceXAI either investing $10 billion into Cursor or acquiring it for $60 billion later this year. A final decision regarding the acquisition or investment has not yet been disclosed.
Currently, Grok 4.5 is unavailable in the European Union, where it is expected to launch in mid-July.
