Elon Musk's SpaceX released Grok 4.5 on Wednesday, its first AI model trained specifically for coding and autonomous agents — a direct challenge to offerings from Anthropic and OpenAI. The launch follows SpaceX's $60 billion acquisition of AI coding startup Cursor, completed just weeks ago, and represents the first tangible product of Musk's vertically integrated AI empire.
SpaceX is not claiming Grok 4.5 leads on benchmark scores. Instead, it is making an economic bet: the model uses half as many tokens per task as comparable models and is priced at $2 per million input tokens and $6 per million output tokens — undercutting rivals by more than half. The company argues developers care more about speed, cost, and practical utility than topping leaderboards.
Grok 4.5 was trained with Cursor, signaling SpaceX's intent to build a closed-loop ecosystem where the model powers the coding tool that helped create it. This vertical integration — combining a launch provider, an AI lab, and a developer platform — gives SpaceX unique advantages in data access and deployment speed that pure-play AI labs may struggle to match.
The pricing strategy could force Anthropic and OpenAI to rethink their own models' cost structures. If Grok 4.5 delivers comparable performance at half the price, enterprise developers may shift workflows, especially for high-volume coding tasks. The broader AI market has seen a price war in recent months, and this latest salvo threatens to compress margins further.
The model is priced at roughly $2 per million input tokens — a significant discount to similar offerings from Anthropic and OpenAI, though exact comparisons depend on task complexity. SpaceX has not released independent benchmark results, leaving questions about how Grok 4.5 performs on complex reasoning tasks versus raw coding throughput.