SpaceXAI Unveils Grok 4.5: A High-Performing Opus-Class AI at a Fraction of the Cost
On a day that will go down in history, coinciding with OpenAI’s GPT-5.6 rollout, SpaceXAI — the AI division of Elon Musk’s recently public Space Exploration Technologies (Nasdaq: SPCX) — has unveiled Grok 4.5 to the public. This marks its most powerful model release to date and the first since the company went public and acquired AI coding startup Cursor.
Grok 4.5 is constructed on a colossal 1.5 trillion-parameter V9 foundation model and enhanced with Cursor’s coding data. This allows it to excel in software engineering, legal analysis, financial tasks, and complex agentic workflows. Elon Musk described it as “an Opus-class model, but faster, more token-efficient and lower cost,” claiming performance on par with Anthropic’s Claude Opus 4.7.
The pricing of Grok 4.5 is a game-changer: it’s available at just $2 per million input tokens and $6 per million output tokens — a stark contrast to Claude Opus 4.8’s $5 input / $25 output. A premium fast-access tier is priced at $4/$18. The model is immediately accessible in Grok Build, all Cursor plans, and the SpaceXAI console, although its EU rollout is delayed.
SpaceXAI has also announced an ambitious roadmap, with plans to release a brand-new foundation model every month through the end of 2026. The model was trained on SpaceXAI’s own Colossus 2 compute cluster in Memphis — the same infrastructure it leases to rivals Anthropic and Google.
Source: TechCrunch — SpaceXAI releases Grok 4.5