Leaked Grok 4.6 Leads in Agent Tasks, Beats GPT-5.6 in Coding

rohanpaul_ai · x · 2026-08-13

Rumors suggest xAI released Grok 4.6. Its reported standout feature is professional agent work, leading GPT Sol Max and Fable 5 Max on GDPVal-AA v2 and AA-Briefcase. It also matches GPT-5.6 at 61 on Artificial Analysis while costing $2/$6 per million tokens. For coding, it purportedly scores 69.9% on CursorBench v3.2, beating Sol's 67.2%. Note: Model names and data are likely unverified or fictional.

Related event: xAI Releases Grok 4.6: Top-Tier Performance at Unmatched Cost Efficiency(77 posts)→

Original post →

More from Fun

Fun channel →