Release2026-07-16

xAI (referred to as SpaceXAI) launched Grok 4.5 on July 16, 2026, positioning it for coding, agentic tasks and knowledge work, and stating it was trained alongside Cursor across tens of thousands of NVIDIA GB300 GPUs with heavy reinforcement learning on multi-step software engineering tasks. Published benchmarks show Grok 4.5 at 62.0% on DeepSWE 1.0, 29.0% on SWE Marathon pass@1 (ahead of Opus 4.8 at 26.0%), 83.3% on Terminal Bench 2.1 and 64.7% on SWE Bench Pro, with competitors Fable and GPT 5.5 leading several of those. The model is served at 80 tokens per second and averages 15,954 output tokens per SWE Bench Pro task, about 4.2x fewer than Opus 4.8. Pricing is $2 per million input tokens and $6 per million output tokens. It is available in Grok Build (as the default model), in Cursor on all plans, and via the xAI API, with free usage in Grok Build and Cursor for a limited time. Word, PowerPoint, Excel and Outlook plugins are offered.

Sources

Send this to someone who needs it

Shares the story and its sources. Nothing about you.

What does this mean for your job?

This is the story as everyone gets it. Once a week we send you the version written for your role — what changed, why it matters for the work you actually do, and one thing to try. Free while we tune it.