Data Science has released a new resource called Model Deployments that allows data scientists and machine learning engineers to deploy models as HTTP endpoints for real time consumption of model predictions. 

• Models have to be saved in the model catalog before they can be deployed through the Model Deployment.
• Model Deployments is available in the OCI Console in a Data Science Project along with Notebook Sessions and Models. 
• Programmatic creation of model deployments and consumption of the model deployment /predict endpoint are possible through the OCI SDKs and OCI CLI. 

If you are using the OCI Python SDK in a notebook session, you must upgrade to the latest version of the OCI Python SDK using:

pip install --upgrade oci 

A few resources to get you started with model deployments are: 

• Model deployments documentation.
• Notebook examples showcasing the end-to-end flow of model training, saving, deploying, and invoking with model deployment.
• For questions/comments, reach out to the Data Science team on our slack channel: #oci_datascience_users