CRITICAL: Project locked out of AI and Support. 404 Model

Google Cloud Build with AI Agents
,
1

Hello, My project (PII Removed by Staff) is completely locked out of using Vertex AI and I am unable to get support. 1. My project has an active PAID billing account (not a free trial). 2. When my Cloud Function tries to call the Vertex AI API (gemini-pro or gemini-1.5-flash) in ANY region (us-central1, europe-west4, etc.), I get the error: “404 Publisher Model… was not found or your project does not have access to it.” 3. My project has the “Basic” support plan, so I cannot create a technical support case. 4. When I try to upgrade my support plan to a paid one (“Standard Support”), the system gives me an error: “You are not eligible to select this option.” I am in a hard lock. I cannot use the AI, and I cannot pay to get help to fix it. This must be an administrative block on my project. Can a Google engineer please investigate project (PII Removed by Staff) and remove this block so I can use Vertex AI? Thank you.

4

Sending you a DM.

1 Like
5

It might be an IAM permissions issue - the service account associated with your Cloud Function needs the roles/aiplatform.user granted on your Google Cloud project. Can you:

  1. Confirm the Cloud Function’s service account has roles/aiplatform.user on the project.
  2. Double check that the Vertex AI API is enabled in your project.
    Let me know if that works!
2 Likes
6

Hola Andrew,

Muchas gracias por tu valioso Feedback.

En efecto, la (IAM) y la verificación de la API de Vertex AI son el trigger sine qua non del reporte.

Sin embargo, es preciso considerar que el proyecto tiene un contexto híbrido que creo que está impactando en su ejecución: el código fuente del proyecto se gestionó en Google Studio, pero el repositorio de código principal reside en GitHub.

Este proyecto se desplegó con las herramientas estándar, pero el error 404 persiste al intentar invocar el modelo desde la Cloud Function.

Con la mayor de mis consideraciones, mi duda es:

¿Hay criterios adicionales de IAM o de autenticación de credenciales externas (Google Cloud Source Repositories vs. GitHub) que puedan estar interfiriendo con la capacidad de la Cuenta de Servicio de la Cloud Function para autenticarse en el endpoint del modelo de Vertex AI?

Agradezco de antemano tu tiempo y atención a la presente.

Mau

7

The fact that your project’s source code is managed in Google Studio but resides on GitHub is unlikely to be the cause of the authentication failure with the Vertex AI endpoint.

The Cloud Function authenticates to other Google Cloud services using its service account, regardless of where the source code was originally stored.

1 Like