SpaceX launched Grok 4.5 on Thursday, which it dubbed its "smartest model yet", also being the first AI model to be trained by AI firm Cursor. The firm stated that the latest in its Grok series of large language models is capable of "excelling at coding, agentic, tasks and knowledge work".
It was trained via datasets consisting of information regarding science, engineering and math while powered by thousands of NVIDIA GB300 Graphical Processing Units (GPUs) .
ALSO READ: AI Is Becoming A Bigger Threat Than Opportunity For Indian IT, Says HSBC
The company claimed that Grok 4.5 can surpass its peers in the realm of engineering tasks, and can create operational applications with a minimum amount of instructions.
According to reports, the AI was able to create a working interactive simulation of the solar system with a single prompt. This capability does not seem to at the cost of speed or expense, as the company claimed that its AI can give faster answers than 'flash' versions of other LLMs and do so for lower costs.
Grok 4.5 costs $2 for every million input tokens and $6 for every million output tokens. In comparison, Open AI's powerful GPT 5.6 model 'Sol' costs $5 per million input tokens and $30 per million output tokens, while its affordable variant Luna, is priced at $1 per million tokens and $6 per million output tokens.
ALSO READ: AI Bulls Bounce Back: Micron, AMD, Intel, Other Chip Hot Bets Rally Up To 9%
Grok 4.5 powers SpaceX's terminal-based AI coding agent Grok build. The model can be accessed across the SpaceXAI console, along with Cursor's plans.
"We've partnered with SpaceXAI to train Grok 4.5. It's our most powerful model yet and the first we've built for more than software engineering," the company said in an 'X' post.
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.