Elon Musk’s SpaceXAI released Grok 4.5 to the public on July 8, the company’s first major model launch since its rebrand from xAI following SpaceX’s acquisition of the company earlier this year, and its first release since acquiring the AI coding startup Cursor.
Musk announced the public rollout directly on X after two weeks of beta testing: “Based on strong positive feedback from customers in our beta test program, @SpaceXAI will make Grok 4.5 available to the public tomorrow. It is an Opus-class model, but faster, more token-efficient and lower cost.” His post is here.
Built for Coding and Agentic Work, Not Chat
Unlike earlier Grok releases pitched primarily as consumer chatbots, Grok 4.5 is positioned first as a coding and agentic-work tool. The model is built on a 1.5-trillion-parameter V9 foundation model with supplemental training on data from Cursor, the code editor SpaceXAI acquired ahead of this release.
It launched simultaneously in Grok Build, inside Cursor across all plans, and through the SpaceXAI API console, with pricing set at $2 per million input tokens and $6 per million output tokens. The model is not yet available to users in the EU.
Musk repeatedly benchmarked the release against Anthropic’s flagship line rather than against OpenAI or Google, calling Grok 4.5 “roughly comparable to Opus 4.7, but much faster,” and arguing that the combination of capability, speed and lower cost — not raw benchmark scores — is what makes it competitive.
He added that the team was “closing the loop on real-world usefulness, not benchmarks,” citing internal use by engineers at Tesla and SpaceX as an early proving ground.
Landing in a Crowded Week
Grok 4.5’s public debut came just ahead of OpenAI’s GPT-5.6 release, in what multiple outlets described as one of the most tightly packed stretches of frontier-model launches to date — SpaceXAI, OpenAI, Google DeepMind and Meta all pushed out major model or platform announcements within roughly the same two-week window in July.
Musk has continued to iterate publicly in the days since launch, posting on X that the model’s context window would expand toward 1 million tokens “probably by next week,” a level of rapid, public iteration that has become a hallmark of SpaceXAI’s release style compared to the more staged rollouts typical of OpenAI, Google and Anthropic.
Independent benchmarking firm Artificial Analysis has since ranked Grok 4.5 fourth on its Intelligence Index, behind Claude, GPT-5.5 and Claude Opus 4.8, though SpaceXAI has emphasized token efficiency and cost — rather than leaderboard position — as its primary competitive pitch to developers evaluating agentic workloads.
Quick Links:
Comments
Be the first to leave a comment.