Technology

SpaceXAI launches Grok 4.5, targets coding and agentic work

Grok 4.5 arrived at $2 per million input tokens, with SpaceXAI pushing it into coding tools, agentic work and broader enterprise tasks.

Marcus Williams··2 min read
Published
Listen to this article0:00 min
SpaceXAI launches Grok 4.5, targets coding and agentic work
AI-generated illustration

SpaceXAI launched Grok 4.5 on Wednesday, July 8, with pricing set at $2 per million input tokens and $6 per million output tokens, a direct bid to win coding and agentic-work users who need models embedded in daily software workflows. The company is pitching the system for software engineering, data science, finance, legal work and other computer-based knowledge tasks, not just casual chatbot use.

SpaceXAI said Grok 4.5 was trained across tens of thousands of Nvidia GB300 graphics processors, with the pipeline built around data filtering, deduplication and quality scoring. That infrastructure-heavy approach points to a model designed for reliability as much as raw capability. The model is already available through Grok Build, Cursor and SpaceXAI’s developer console through an API key, and European availability is expected in mid-July.

Cursor said Grok 4.5 was trained jointly with SpaceXAI on a new pretrained mixture-of-experts model and then continued on trillions of tokens of Cursor data. Cursor also said it added new cybersecurity safeguards tied to the model’s capabilities, and it is offering double usage for the first week to individual and team subscribers. The rollout puts Grok 4.5 directly inside one of the most widely used coding environments, where product integration can matter as much as model quality.

The pricing sharpens the contest. At $6 per million output tokens, Grok 4.5 undercuts Anthropic’s Claude Opus 4.8 on the output side, while Elon Musk described the model on X as faster, more token-efficient and lower cost than an Opus-class rival. For enterprises weighing agentic tools, the real comparison is no longer just benchmark scores. It is whether a model can hold up under security controls, fit into existing developer workflows and stay economical at scale.

The launch also lands amid broader corporate maneuvering around Cursor and enterprise AI tools. SpaceX said in June it would acquire Anysphere, the startup behind Cursor, in a $60 billion all-stock deal. Musk said in May that xAI would cease to exist separately and instead become SpaceXAI after an earlier acquisition by SpaceX. TechCrunch called Grok 4.5 the company’s first major model release since it went public several weeks ago, while OpenAI was preparing a public rollout of GPT-5.6 on Thursday, adding another layer to a market fight now being waged on price, integration and trust.

This article was produced by Prism’s automated news system from verified source data, official records, and press releases, then run through automated quality and moderation checks before publishing. The system is built and supervised by the people who set the standards it runs under. Read our full AI policy.

Did this article answer your question?

Discussion

More in Technology