Managed model deployments

Managed deployments let an entitled organization member launch a RunPod Serverless endpoint without entering provider credentials or arbitrary infrastructure settings. The feature must be enabled and configured by a platform administrator.

Launch a deployment

1. Open /model-deployments and choose a qualified profile.
2. Review the organization quota and estimated provider configuration.
3. Launch the deployment and follow its provisioning status.
4. Wait for the endpoint and expected model to pass readiness and qualification.
5. Select the model in a Thread only after the deployment reports ready.

The deployment remains unavailable while provisioning, validation, or model qualification is incomplete. Members cannot supply arbitrary images, credentials, or raw provider fields through this flow.