Vertex AI model registry

Google Cloud Build with AI Custom ML & MLOps
,
1

I am using Vertex AI Registry for my organization. I have created a project where we send all our models. After the models are created, I want to create an endpoint in Project B and deploy the сustom-model from Project A, where the models are stored. I am using the gcloud command and getting an error that the model does not exist.

Projects A and B are in the same region. Am I correct in understanding that I cannot use a model from Project A for deployment in an endpoint in Project B, and that I need to either copy the model or register it in Project B?

I am successfully able to copy models between projects. I have given the service account from Project A in Project B the “Owner” and “Platform Admin” roles, and I did the same for the service account from Project B in Project A. I can successfully deploy a model to an endpoint if the model is in the same project.

Command:

gcloud ai endpoints deploy-model 1141570146561687552 \
  --project=(PII Removed by Staff) \
  --region=northamerica-northeast1 \
  --model=projects/(PII Removed by Staff)/locations/northamerica-northeast1/models/2006322847667978240 \
  --display-name="my-endpoint" \
  --machine-type=n1-standard-4 \
  --traffic-split="0=100"
(gcloud.ai.endpoints.deploy-model) NOT_FOUND: Model
2

Hello @Alik_Mkrtchian,

I think that you only need to pass the model ID without other information.

From gcloud ai endpoints deploy-model documentation :

EXAMPLES

To deploy a model 456 to an endpoint 123 under project example in region us-central1, run:

gcloud ai endpoints deploy-model 123 \
--project=example \
--region=us-central1 \
--model=456 \
--display-name=my_deployed_model