Once we have published our updated "predictor-only" editquality model-server image to the WMF Docker Registry and have uploaded all editquality model files to storage, we can add the Inference Service configurations to the ml-services helmfile.
We can follow the Deployment guide that @elukey put together to do this:
https://wikitech.wikimedia.org/wiki/Machine_Learning/LiftWing/Deploy#Add_a_model_to_an_existing_helmfile_config
We mostly need to update the editquality image version to the new predictor-only image and then add an entry under inference_services for each model we want to deploy.