SpaceXAI released Grok 4.5 on Wednesday. It’s the first model since the company went public, and they’re positioning it as a workhorse — coding, research, writing, office work. All the typical stuff.
The big claim here is cost. SpaceXAI says Grok 4.5 has “twice greater token efficiency” than other leading models. At $2 per million input tokens and $6 per million output, it’s much cheaper than the competition. Anthropic’s Opus 4.7 runs $5 input and $25 output. OpenAI’s Sol costs $5 input and $30 output.
Elon Musk called it an “Opus-class model” on X — faster, more token-efficient, and lower cost. He later clarified their internal assessment puts it roughly on par with Opus 4.7 but significantly faster.
Benchmarks show Grok 4.5 competing with top models, though not quite best-in-class across the board. That’s still impressive for the price point.
The timing’s interesting. OpenAI plans to release GPT 5.6 (codenamed Sol) on Thursday — their most powerful model yet, though its release was previously held up by the Trump administration over security concerns.
We’re in the middle of a big AI model dump. Grok 4.5 today, GPT 5.6 tomorrow. The pricing war is heating up, and that’s good for anyone using these tools.
