+
K
A deployed instance of a model that can serve live inference requests. Live deployments manage the compute resources needed to host a model version and expose it for real-time inference via the transformJson operation.