SAN FRANCISCO — xAI is about to raise the tempo again. Elon Musk says the company's next model, Grok 4.6, will launch around August 7, with an even larger Grok 4.7 to follow weeks later — a back-to-back release cadence that no other frontier AI lab is currently matching in public.
Musk laid out the timeline in a post on X, describing Grok 4.6 as a 1.5-trillion-parameter model with significantly improved supervised fine-tuning and reinforcement learning built on the Grok 4.5 foundation. Grok 4.7, he said, will be a larger 2.1-trillion-parameter model that is "better than 4.6 in every way, except slightly slower to serve, albeit with even better token efficiency."
Speed as a Strategy
What stands out is the pace. Grok 4.5 launched in early July, and within days Musk was already confirming its successor had finished initial training. xAI is shipping updates faster than OpenAI or Anthropic are managing publicly — a tempo that turns release velocity itself into a competitive weapon. The company has been pushing Grok deeper into everyday products, too, from the Grok Imagine tool that just added native 1080p video generation to faster on-device experiences.
Musk's pitch is that smarter architecture beats brute-force scale. Rather than chase raw parameter counts, he says Grok 4.6 aims to match or exceed the capability of China's Kimi K3 while keeping the generation speed and token efficiency of the smaller Grok 4.5 — an important edge given how expensive it is to run enormous models in production.
Taking Aim at Kimi K3
The competitive target is explicit. Moonshot AI's Kimi K3, a roughly 2.8-trillion-parameter open-weight model released in July, is the largest of its kind and has pressured U.S. labs that mainly distribute their most advanced models through closed platforms. As crypto.news reported, Moonshot even welcomed xAI to the "2 trillion-plus club" in a good-natured jab, underscoring a rivalry that has become genuinely public.
xAI enters that contest from a position of strength on price. In a recent independent assessment, Grok 4.5 was rated the best-performing model relative to its cost — roughly 10 times cheaper than one leading U.S. frontier model and more than twice as cheap as Kimi K3 while delivering comparable performance. If Grok 4.6 preserves that efficiency at a higher capability tier, xAI could extend its price-performance lead just as demand for frontier models accelerates. Details on Grok 4.6's benchmarks and pricing remain on xAI's official news channel ahead of launch.
The Broader Push
The rapid cadence reflects xAI's expanding footprint across Musk's empire, where Grok now powers experiences from cars to the X platform and, increasingly, enterprise workflows. Each release also feeds a virtuous cycle: more capable, cheaper models attract more usage, which in turn funds the compute build-out — including the orbital ambitions SpaceX is pursuing for AI infrastructure.
With Grok 4.6 due within days and Grok 4.7 already in the pipeline behind it, xAI is signaling that it intends to keep the pressure on rivals through sheer momentum. For a company that did not exist three years ago, the message is clear: in the AI race, xAI plans to lead not by waiting, but by shipping.
