Grok 4.7 achieves frontier price-performance on CursorBench 4.0
September 21, 2026
Grok 4.7 utilizes a larger base model trained with extended reinforcement learning on long-duration tasks. The model delivers twice the speed and half the cost per task compared to previous versions, specifically targeting high-complexity coding and knowledge work.
HOW THIS AFFECTS YOU
●
builderYou can deploy more complex coding agents at 50% lower cost and higher throughput.
●
founderYou can improve your product's margins by switching to a more efficient frontier model for long-running tasks.