Grok 4.5 is available now: cheaper pricing and coding tools

xAI says Grok 4.5 is out now, faster responses, lower operating costs, and a stronger push into coding, knowledge work, and agent-style tasks.

Pricing starts at $2 per million input tokens and $6 per million output tokens. xAI also says the model can complete tasks in less than half the steps, which could bring down your real costs on repeat workloads.

You’re also getting a model meant to handle software engineering and day-to-day office work. xAI says Grok 4.5 uses far fewer output tokens than Anthropic’s Opus 4.8 on SWE Bench Pro, can match GPT-5.5 in some coding-agent tests, and is built to help with spreadsheet models, presentation decks, and document drafting. Part of that comes from training on tens of thousands of Nvidia GB300 GPUs and data from Cursor.

If you want one model that can cover both development and productivity work, this launch looks like one to watch. Early reaction has centered on speed, efficiency, and how useful it feels in practice, and Artificial Analysis has already placed Grok 4.5 on the performance-versus-cost Pareto frontier, even if some rivals may still be ahead on raw capability.

Grok 4.5 is available now.