Elon Musks SpaceXAI Launches Grok 4.5 for Coding and Agentic Tasks

TLDR

• SpaceXAI launched Grok 4.5 for coding and agentic tasks.
• Grok 4.5 is available through Cursor and Grok Build.
• SpaceXAI priced Grok 4.5 at $2 per million input tokens.
• The model was trained on Nvidia GB300 GPUs and Cursor data.
• Elon Musk called Grok 4.5 an “Opus-class model.”

SpaceXAI has launched Grok 4.5, a new AI model built for coding, software engineering, and long-running agentic tasks, as Elon Musk’s company expands deeper into enterprise AI tools.

The company said Grok 4.5 is its most intelligent model to date and is available through Grok Build, Cursor, and the SpaceXAI developer console. EU availability is expected in mid-July, while pricing starts at $2 per million input tokens and $6 per million output tokens.

Grok 4.5 Targets Coding and Agentic Work

SpaceXAI said Grok 4.5 was trained across tens of thousands of Nvidia GB300 GPUs. The company said training focused on data filtering, deduplication, quality scoring, and domain selection to improve performance in coding and real-world software tasks.

The model is designed to handle long tasks that require tool use, planning, and work across large codebases. Cursor said, “We’ve partnered with SpaceXAI to train Grok 4.5.” The AI coding agent also said the model can support software engineering, data science, finance, legal work, and other computer-based tasks.

Cursor described Grok 4.5 as a mixture-of-experts model trained with trillions of tokens of Cursor data. That data included user interactions with codebases and software tools, allowing the model to learn from both existing software and developer-agent workflows.

SpaceXAI said the model performs well on coding tasks involving Rust, C, C++, and full app development from prompt to production. The company also said Grok 4.5 can work across multiple repositories and tools.

Musk Says Model Is Lower Cost Than Opus

SpaceX CEO Elon Musk said on X, “It is an Opus-class model, but faster, more token-efficient and lower cost.” The statement compared Grok 4.5 with high-end models used for complex reasoning and coding tasks.

Grok 4.5 is priced below Anthropic’s Claude Opus 4.8, which costs $5 per million input tokens and $25 per million output tokens. OpenAI’s GPT-5.6 Luna is priced at $1 per million input tokens and $6 per million output tokens.

SpaceXAI also offers a faster Grok 4.5 variant priced at $4 per million input tokens and $18 per million output tokens. Individual and team plans include usage through the company’s first-party model pool, with doubled usage offered during the first week.

Input tokens refer to text, code, or other data sent to an AI model. Output tokens are the text, code, or other responses generated by the model.

Cursor Deal Supports SpaceXAI Expansion

SpaceX said last month it would buy Anysphere, the company behind Cursor, in an all-stock deal worth $60 billion. The acquisition is aimed at strengthening SpaceXAI’s position in enterprise AI coding tools.

Musk’s AI startup xAI was acquired by SpaceX in February. Musk later said xAI would no longer operate as a separate company and would become SpaceXAI.

Grok 4.5 is now available through Cursor across desktop, web, iOS, CLI, and SDK access. Developers can also use the model through the SpaceXAI console with an API key.

OpenAI is expected to launch GPT-5.6 publicly on Thursday after a delay tied to US government requests over national security concerns.

The post Elon Musks SpaceXAI Launches Grok 4.5 for Coding and Agentic Tasks appeared first on CoinCentral.