On 21 August 2026, Elon Musk posted on X:
Grok 4.6 on extra high thinking mode now achieves #1 score on CursorBench!
The post quoted @XFreeze, who published a CursorBench 3.2 comparison that listed Grok 4.6 Extra High first, alongside other named models and per-task cost figures.
What was stated
- Elon attributed the #1 CursorBench score to Grok 4.6 in extra high thinking mode
- The quoted post described the leaderboard as CursorBench 3.2
Grokati is reporting Elon’s statement and the existence of that quoted comparison. It is not an independent benchmark run.
Limits of this report
No verification of CursorBench methodology, the cost-per-task figures, or later leaderboard changes is included. Those numbers belong to the quoted post, not to an xAI product announcement.
Source
Elon Musk on X, 21 August 2026.