Grok 4.6 just ranked #
1# on CursorBench 3.2
Outperforming Claude Fable 5, Opus 5 and GPT-5.6 Sol on real-world coding performance
And what makes this even crazier is the efficiency....the chart gives CursorBench performance against average cost per task, and Grok 4.6 is sitting right at the top of the frontier
Grok 4.6 is insanely capable at coding....delivering frontier coding performance at ridiculous efficiency