SAN FRANCISCO — SpaceXAI rolled out Grok 4.5 on Wednesday, the most capable version yet of its Grok model family and the first major release since Elon Musk folded xAI into SpaceX earlier this year. The company is positioning it as a workhorse built for real work — coding, research, and office tasks — at a fraction of the cost of rival systems.
It is the clearest sign yet that Musk's AI ambitions and SpaceX's resources are now pulling in the same direction, backed by the same rockets, capital, and Colossus supercomputer that power the rest of the empire.
An 'Opus-Class' Model
Musk described Grok 4.5 as "an Opus-class model, but faster, more token-efficient and lower cost," a direct comparison to Anthropic's flagship Opus family. He added that the company's internal assessment puts Grok 4.5 "roughly comparable to Opus 4.7, but much faster," and framed the blend of capability, speed, and price as its main advantage.
The model was trained jointly with Cursor, the AI coding platform SpaceX agreed to acquire in a $60 billion stock deal, and it was fed trillions of tokens capturing how developers actually interact with codebases and software tools. That focus shows up in the benchmarks: SpaceXAI says Grok 4.5 leads on several coding and terminal tests while trailing only the very largest models elsewhere.
Priced to Undercut Rivals
Cost is central to the pitch. In its announcement, SpaceXAI priced Grok 4.5 at $2 per million input tokens and $6 per million output tokens — well below Opus 4.7's $5 and $25. The company also claims "twice greater token efficiency" than other leading models, a meaningful edge as businesses watch their AI spending more closely.
Grok 4.5 targets not just coders but legal and financial teams, with added cybersecurity features aimed squarely at enterprise customers. It is available now in Grok Build, in Cursor across all plans, and through the SpaceXAI console, with a broader public release following this week.
Building on a Fast-Moving Stack
The launch caps a rapid stretch for the newly combined company, which has been steadily expanding Grok's developer stack with voice, agent, and tooling features. Each release has pushed Grok further from a novelty chatbot and closer to a full productivity engine.
With Grok 4.5, Musk is betting that a cheaper, coder-friendly model can win paying business even while it chases the very top of the leaderboard. If the efficiency claims hold up in real-world use, SpaceXAI has handed developers and enterprises a compelling new option — and signaled that the cadence of Grok releases is only accelerating, with Musk already promising to close the remaining gap to the largest frontier models "soon."
