Vertex AI - "gcloud ai models list" returns 0 items on a fully enabled and activated projec

Google Cloud Build with AI Generative AI & Foundational Models
1

Hello, I am unable to list or access any Vertex AI models (like Gemini) in my project, (PII Removed by Staff). My gcloud command returns "Listed 0 items," and my Node.js application gets a 404 NOT_FOUND error. I believe this is a backend provisioning issue with my project.

Then, provide all the diagnostic information and troubleshooting steps you have already taken. This is the most important part, as it shows you have done your due diligence. Include:

  • Project ID: (PII Removed by Staff)

  • The Audit Log: Paste the full JSON from the audit log that showed the Resource ... was not found error.

  • Troubleshooting Steps: List everything you have already confirmed:

    • The Vertex AI API is Enabled.

    • The project has an Active Billing Account (upgraded from the Free Trial).

    • Your user (PII Removed by Staff) has the Vertex AI User IAM role.

    • You have refreshed your gcloud authentication.

    • You have confirmed there are no restrictive Organization Policies.

1 Like
2

appears changing this to gemini-2.5-flash is now working but its running slower, but is working

3

Short answer: gcloud ai models list only shows models in your project (custom models you trained/imported). It will not list Google’s publisher (Model Garden) models like Gemini — that’s why you see Listed 0 items..

Quick, actionable checklist to diagnose & fix:

  1. Confirm Model Garden visibility in the Console — go to Vertex AI → Model Garden and check whether Gemini (or other publisher models) appears for your project.

  2. Test publisher-model access with a direct API call (replace PROJECT_ID, LOCATION, MODEL_ID):

ACCESS_TOKEN=$(gcloud auth print-access-token)
curl -X POST \
  -H "Authorization: Bearer $ACCESS_TOKEN" \
  -H "Content-Type: application/json" \
  "https://${LOCATION}-aiplatform.googleapis.com/v1/projects/${PROJECT_ID}/locations/${LOCATION}/publishers/google/models/${MODEL_ID}:generateContent" \
  -d '{"contents":[{"role":"user","parts":[{"text":"Hello"}]}]}'

If that returns text, your project can reach the publisher model.
3. Double-check the model name/version (Gemini 1.x was deprecated; use current Gemini 2.x/2.5 IDs) and make sure your client/SDK/gcloud are up to date — older names or SDKs can produce 404s.
4. If the console doesn’t show the model and the API call 404s, verify Model Garden access controls / org policies (some orgs restrict publisher-model access) or open a GCP support case — 404 often means your project doesn’t have access/provisioning for that publisher model.

1 Like