Grok 4.6 arrives for long-running agents

Grok 4.6 looks less like a chatbot refresh and more like xAI’s push into serious agent infrastructure.

• –500k context makes it suitable for large codebases and extended research workflows
• –Availability in Cursor and Grok Build puts the model directly inside developer tools
• –Pricing starts at $2 per million input tokens and $6 per million output tokens
• –xAI’s benchmark claims are promising, but independent evaluations should determine real-world reliability
• –Configurable reasoning from low to xhigh gives developers a useful cost-versus-quality control