Mlflow and MlflowClient differences in model registry

I noticed that while trying some code: There’re 2 ways to create a containing registry model

1. using mlflow directly by registering an ML model and specify the containing model this method will create the registry model if not exist then create the version for the MLmodel

mlflow.register_mode(ml_model_uri, name_of_the_containing_model)

2. using MlflowClient object

client = MlflowClient(backend_store_uri)
client.create_registred_model(params)
client.create_model_version(params) 

The create model version will not automatically create the containing model so you must create the registered model first using create_registered_model