Grok 4.7 beats GPT-6 Astra on professional work, leading by 153 Elo on GDPval-AA

XFreeze · x · 2026-09-22

On the same Artificial Analysis run, Grok 4.7 leads GPT-6 Astra by +153 Elo on GDPval-AA (professional deliverables) and +88 Elo on AA-Briefcase (multi-hour office work), putting it ahead on both professional output and long-form knowledge work. The poster says SpaceXAI is pushing Grok hard into real-world knowledge work.

Related event: Grok 4.7 Enters Top Four on Intelligence Index at Half of Opus 5's Cost, but Token Use Doubles(16 posts)→

Original post →

More from Models

Models channel →