SpaceXAI (the AI company formerly known as xAI) today released Grok 4.6, its latest flagship model, emphasizing stronger performance on long-running agents and more ambitious interactive and visual tasks. The update continues the rapid iteration cycle that began with Grok 4.5 earlier this summer.
Hours after the launch, Elon Musk posted on X that Grok 4.7 is already significantly better than 4.6 and is expected in three to four weeks. Initial training is complete; the team is now running supplemental training on a massive amount of SpaceX company data. Musk described the upcoming model as “something special” and said it “will exceed all current models."
He specifically highlighted the unique value of the SpaceX dataset: “the SpaceX training corpus is so awesome & unique that I would be shocked if any model is better at real-world engineering than 4.7.” While acknowledging that Anthropic is a strong competitor likely to release improved models soon, Musk positioned the SpaceX data as giving Grok 4.7 a distinctive real-world engineering advantage.
Grok 4.6 itself is positioned as a clear step up from 4.5 at the same price point, with particular gains in staying on complex multi-step tasks—whether research, codebase work, or turning ideas into polished applications. Early availability includes Cursor, Grok Build, the API, and select partners, with temporary double usage allowances in some products.
The rapid succession of releases and the planned injection of proprietary aerospace and engineering data have generated notable excitement about potential breakthroughs in aerospace-related problem-solving and other complex real-world domains.
Have a correction or tip? See our
corrections policy or contact the newsroom.
