SpaceXAI launched Grok 4.6 and said Grok 4.6 High scored 61 on the Artificial Analysis Intelligence Index. That ties it with GPT-5.6 Sol Max on the same index, while putting the new model into Cursor, Grok Build, the API, OpenRouter, Vercel, and Cloudflare.
Grok 4.6 High and 61
The company said Grok 4.6 improves multi-step task execution, complex programming, and knowledge work. It also said the model went through a longer period of supplementary training, with data covering reasoning, advanced technical concepts, and high-quality engineering content.
SpaceXAI said the later training stage covered knowledge work, general programming, kernel optimization, web development, and computer-aided design. Those are the parts of the stack that matter to users who push models through long tasks rather than short prompts.
Cursor and Grok Build quotas
Users of Cursor and Grok Build got double usage quotas during Grok 4.6's first week. That makes the launch less like a lab release and more like a live rollout for people already running AI tools inside daily workflows.
The standard version costs $2 per million input tokens and $6 per million output tokens. The high-speed version costs twice the price of the standard version, which leaves buyers with a simple tradeoff between throughput and spend.
GPT-5.6 Sol Max comparison
The friction point is the gap between the benchmark headline and the price sheet. Grok 4.6 High is being presented as equal to GPT-5.6 Sol Max on the Artificial Analysis Intelligence Index, but the standard model is priced low enough to invite testing in real workloads before anyone treats the score as the whole story.
SpaceXAI has not said when the first-week quota promotion ends after that opening window, so the practical move for users is to check their own limits and route the model through the platform they already use most.
