Hello,
I am trying to create a chatbot.
I am trying to call Vertex AI foundation models (Gemini) from Cloud Run (and also tested locally).
I have created a brand new project (fif-chatbot, ID 71350167616) specifically for this.
I have verified Billing is active and the account is in good standing.
I have verified the Vertex AI API is enabled.
I have verified the default compute service account has Editor, Vertex AI User, and Service Usage Consumer roles.
I consistently receive the error FailedPrecondition: 400 Project 71350167616 is not allowed to use Publisher Model ‘projects/fif-chatbot/locations/us-central1/publishers/google/models/gemini-1.0-pro-001’ when the application tries to call model.generate_content().
This happens even for the generally available gemini-1.0-pro-001 model.
I tested this in multiple regions (us-central1, europe-west9) with the same result.
I tested both from Cloud Run (using ADC) and locally (using a Service Account Key) with the same result.
I am on the Basic Support plan and cannot file a direct technical case.
Is there might be a project provisioning issue, a hidden Terms of Service acceptance required, or any known issues preventing new projects from accessing foundation models.
Hope someone can help. Thanks.