SpaceXAI announces Grok 4.5, an "Opus-class" model cheaper than rivals

The announcement and the core numbers

Elon Musk announced Grok 4.5 on July 8, 2026, the first model release from SpaceXAI (the name xAI adopted after merging with SpaceX) since the rebrand. The company describes Grok 4.5 as a model comparable in class to Claude Opus 4.7, but faster and cheaper to run. Musk's own framing placed the model directly against Anthropic's flagship line, rather than against cheaper, mid-tier competitors. According to TechCrunch, public availability of the model is expected on July 9. The announced pricing is $2 per million input tokens and $6 per million output tokens. xAI stated that coding, agentic tasks and office work are the model's main focus areas.

Price comparison with the competition

Grok 4.5's price sits well below its direct competitors. Anthropic's Claude Opus 4.7 charges $5 per million input tokens and $25 per million output tokens. OpenAI's GPT-5.6 Sol charges $5 for input and $30 for output per million tokens, according to figures cited by Bloomingbit. In practice, Grok 4.5 comes out cheaper on both ends, and especially on output: less than a quarter of GPT-5.6 Sol's output price, and less than a quarter of Opus 4.7's as well. That gap matters because output tokens, not input tokens, tend to dominate the bill in longer, more complex responses.

What changes for those building or reselling AI products

For small and medium businesses and agencies building customer service agents, marketing automation and AI-driven workflows, the direct effect of a cheaper frontier model is lower cost to run those agents in production. Agentic workloads tend to generate large volumes of output tokens, such as intermediate reasoning, tool calls and revisions, and that is exactly where Grok 4.5 looks most competitive. That opens room for operations that previously relied on cheaper, less capable models to move to an Opus-class model without multiplying the cost per interaction. In practical terms, an agency running thousands of support or sales conversations a day through an AI agent could adopt a top-tier model while keeping its per-conversation cost closer to what it used to pay for a weaker one.

What this price war signals

The contest between Opus 4.7, GPT-5.6 Sol and Grok 4.5 shows that the capability frontier and the price frontier no longer move together. Having the most capable model is no longer enough: the race is now also about who delivers that level of capability at the lowest cost per token. If this pattern holds, it should accelerate adoption of frontier models in high-volume applications, precisely the kind of use case still limited by cost today. For buyers, that likely means faster price drops across the board as each lab tries to avoid being the most expensive option at a comparable capability level.