The model for long agentic tasks is now in the Copilot picker - VS Code, CLI and cloud agents. Some business accounts have to enable it in settings. In the API it is $2/$6 per million tokens.
- On 14 August SpaceXAI announced Grok 4.6 in GitHub Copilot: VS Code, Copilot CLI and cloud agents, via the model picker.
- For some business and enterprise accounts the model has to be enabled from Copilot settings.
- In the SpaceXAI console: $2 per million input and $6 per million output tokens; a fast variant at twice the price.
Two days passed between the model's launch and the Copilot picker, and that gap says more about the market than the model itself.
Look at what is happening to the Copilot picker. A year ago you chose between two or three models. Today every big maker wants a line there, because that is where millions of people actually write code, and whoever is not in the picker might as well not exist for them.
For you: try it on one real task, not a demo. And check who pays - Copilot or your API key.