Grok 4.6 Just Landed on Google's Enterprise Agent Platform, With a 500K-Token Context Window

xAI has added Grok 4.6 to Google’s Enterprise Agent Platform through Model Garden, joining xAI’s own API, Cursor, Grok Build, OpenRouter, Vercel, and Cloudflare as distribution channels for the model. The release ships with a 500,000-token context window and configurable reasoning levels — low, medium, high, and xhigh — explicitly positioned for long-running agentic workflows rather than single-turn chat. Google’s platform pricing for Grok 4.6 runs $2 per million input tokens, $0.50 per million cached input tokens, and $6 per million output tokens.

The distribution strategy matters as much as the model itself. xAI has spent the past several weeks embedding Grok into wherever enterprise workflows already happen rather than pulling users into a standalone product: Grok Bot expanded to SuperGrok Plus, SuperGrok Heavy, and Cursor Pro+/Ultra/Teams plans in beta without introducing new pricing tiers, and before that, xAI shipped a free Grok add-on for Google Workspace — Sheets, Slides, Docs — following the same embedded-in-the-productivity-suite playbook it had already run in Microsoft 365 across Word, Excel, PowerPoint, and Outlook. Landing on Google’s own Enterprise Agent Platform, alongside whatever first-party Gemini models Google offers there, is the next logical extension of that same strategy: compete for enterprise agent workloads by being available inside the platforms enterprises already standardized on, rather than requiring them to switch.

For technology leaders evaluating model providers for agentic use cases, the practical signal is availability, not just capability: a 500K-token context window is a meaningful spec on its own, but a model that shows up natively inside Google’s, Microsoft’s, and Cursor’s existing enterprise tooling lowers the actual switching cost of testing it against whatever’s already running in production — which is increasingly the deciding factor over benchmark scores alone.