Grok 4.5 Launched

1.5 trillion parameters marks the massive scale of SpaceX AI's newly released Grok 4.5, a model that is 3 times larger than its predecessor, Grok 4.3. This massive jump in size represents a pivotal shift for the lab, as they publicly launched the system as their very first model trained specifically for coding and agents.

To bring this to life, SpaceX AI partnered with the development platform Cursor to train what Cursor calls their most powerful model yet, expanding its capabilities far beyond basic software engineering. The launch was strategically timed to land just a day before the confirmed release of OpenAI's GPT-5.6, allowing SpaceX AI to capture the spotlight first.

According to evaluations by Artificial Analysis, Grok 4.5 has secured the number 4 spot on their intelligence index with a score of 54, placing it right behind the leading models from OpenAI and Anthropic. While Elon Musk framed the model as roughly comparable to Anthropic's Opus 4.7 but much faster, analysts at Artificial Analysis suggest it sits on the Pareto frontier for cost and performance.

This efficiency is highly visible in the pricing, with Grok 4.5 costing just $2 per million input tokens and $6 per million output tokens, compared to $5 input and $30 output for the upcoming GPT-5.6. This aggressive pricing structure positions SpaceX AI to challenge the flagship coding agent tier by offering near-frontier capabilities at a fraction of the cost of its rivals.