Skip to main content
When you run a session, data required for model execution is sent to the provider/harness you selected. Practical rule:
  • If a run needs model inference, relevant prompt/context is processed by that configured provider path.
Use your deployment’s settings and policies to control:
  • which providers are enabled
  • which runtime/network modes are used
  • telemetry behavior
For organization-specific policy guarantees, follow your internal security guidance for ctx deployment.