SpaceXAI just dropped Grok 4.5, and the AI pricing war has a new frontrunner. The model, launched on July 8, costs $2 per million input tokens and $6 per million output tokens, making it over 60% cheaper than both Anthropic’s Claude Opus 4.8 and OpenAI’s GPT-5.5.
What Grok 4.5 actually brings to the table
The model features a 500K token context window and operates at roughly 80 tokens per second. In benchmark testing, it achieves 2 to 4.2 times better token efficiency compared to Opus 4.8 and GPT-5.5 on equivalent tasks, meaning it does the same work while burning through far fewer tokens, which directly translates to lower costs for developers and businesses.
Grok 4.5 ranked first or near the top on several key benchmarks, including AutomationBench-AA and various legal agent evaluations. It currently sits at number four on the Artificial Analysis Intelligence Index, behind Fable 5, GPT-5.5, and Opus 4.8.
The model was trained on trillions of tokens, pulling heavily from Cursor’s coding datasets and running on tens of thousands of NVIDIA GB300 GPUs. It’s available immediately through Grok Build (where it’s the default model), Cursor across all plans, and the SpaceXAI API. An EU rollout is expected by mid-July.
