SpaceX's New AI Model Grok 4.5: What It Does and Why the Price Matters

SpaceX released a new artificial intelligence model called Grok 4.5 on July 8, 2026. Elon Musk said it was as smart as Anthropic's best model, called Opus 4.7, but runs faster TechCrunch. The model became publicly available the next day.

How Much Does It Cost?

The price is lower than competitors. SpaceX charges $2 per million input tokens and $6 per million output tokens. Anthropic charges $5 and $25 for the same work TechCrunch.

To understand why this matters, think of tokens like words. When you ask an AI a question, you pay for the words you put in and the words it gives back. If an AI costs less per word and produces the same quality answers, businesses will use it more often because it saves money.

SpaceX says Grok 4.5 needs fewer tokens to do the same job as other models — meaning it wastes less. The company is also offering a free trial at launch to get people to try it.

What Is It Built For?

SpaceX built Grok 4.5 specifically for three things: writing and checking code, handling tasks that require an AI to think through problems step-by-step, and helping with work that involves lots of documents and reading. SpaceX partnered with Cursor, a company that makes coding tools, to market the model for these uses. The company also plans to push it into legal and finance work, where reading through contracts and documents matters most.

Before the public release, SpaceX tested Grok 4.5 inside its own companies — SpaceX and Tesla — to make sure it worked well on real problems, not just test scores.

Who Is Investing in SpaceX?

SpaceX raised $6 billion in funding for its AI division. The money came from a mix of venture capital firms like Andreessen Horowitz and Sequoia Capital, but also from giant investment companies and wealth funds from other countries. BlackRock and Fidelity — two of the largest investment firms in the world — are named as investors. This tells us that major financial institutions now see AI models as a serious business, not just a gamble.

Why Did This Release Happen Now?

Musk said SpaceX released Grok 4.5 because customers in its testing program gave positive feedback. OpenAI announced it would release a new model, GPT 5.6, one day later, on July 9. Neither company has said whether the timing was planned as a competitive response, though SpaceX's public statements focus on customer feedback rather than beating a rival to market.

The Bigger Picture

The name "Opus-class" — which Musk used to describe Grok 4.5 — has become an informal way the industry talks about AI models at a certain level of smartness. A few years ago, "GPT-4-class" was the benchmark. The fact that SpaceX is now comparing its model to Anthropic's instead of OpenAI's shows how the competition among major AI labs has shifted.

SpaceX also owns X, the social platform formerly known as Twitter. This means SpaceX can reach customers directly through X in ways that OpenAI and Anthropic cannot. This is a real competitive advantage.

The pricing difference is significant, especially for work where the AI generates a lot of output — like thinking through a complex problem step by step. Whether SpaceX can keep these prices this low when lots of customers start using it is an open question. History shows that vendor-announced price differences often change once real-world usage starts. The claimed efficiency improvements also deserve independent testing before people rely on them fully. These are details worth watching as more users try the model.