Grok 4.6 (xAI)

Grok 4.6 is positioned as a frontier successor to Grok 4.5 from xAI, targeting direct competition with Moonshot's Kimi K3 and Claude Opus 4.8. According to third-party reporting, the model reuses the same 1.5-trillion-parameter V9 foundation as its predecessor rather than scaling up raw size. The reported improvement comes from substantially enhanced supervised fine-tuning (SFT) and reinforcement learning (RL) pipelines, with xAI betting that stronger post-training can match or exceed much larger competitors at roughly half the weight of Kimi K3.

Practically, Grok 4.6 is intended to preserve the speed and token efficiency that defined Grok 4.5, reportedly holding the same throughput envelope of around 80 transactions per second. Distribution spans the xAI API, the Grok app, grok.com, and SuperGrok / X Premium+ tiers, with third-party evaluation surfaces such as arena.ai also cited. A larger 2.1-trillion-parameter Grok 4.7 is reportedly planned for a few weeks later as a step-up, while Grok 4.6 itself fits teams needing frontier-grade reasoning throughput without waiting for the bigger release.