Gemini API generateContent returns 403 while models/countTokens and quotas work

Gemini API
1

Hi Gemini API team,

I’m hoping someone can help me check whether our project is stuck in a project-level access or quota-tier state.

We are trying to use the Gemini Developer API from an organization-owned Google Cloud project for a production app. The project is visible in Google Cloud Console and AI Studio, and the Gemini API itself appears to be enabled, but every GenerateContent request is denied.

Project details:

  • Project ID: petlog-gemini-probe-2026061..
  • Project number: 6077702270..
  • Google Cloud organization: petzlogom
  • API: generativelanguage.googleapis.com
  • Current primary model: gemini-3.1-flash-lite
  • Runtime key type: service-account-bound API key, restricted to the Generative Language API
  • Billing state: free-tier project, no paid billing attached

What works:

  • `GET /v1beta/models` returns HTTP 200 and lists the expected models.
  • `countTokens` returns HTTP 200 for the tested models.
  • Cloud Console > Gemini API > Quotas & System Limits shows non-zero free-tier quota rows.
  • For example, `gemini-3.1-flash-lite` shows 15 RPM, 500 RPD, 250,000 input tokens per minute, and Unlimited input tokens per day.
  • Current usage is below those limits.
  • AI Studio can see the project after importing it.

What fails:

  • `generateContent` returns HTTP 403 for every tested model.
  • The error is:
    `PERMISSION_DENIED: Your project has been denied access. Please contact support.`
  • This happens with `gemini-3.1-flash-lite`, `gemini-3.5-flash`, `gemini-2.5-flash-lite`, `gemini-2.5-flash`, `gemma-4-31b-it`, and `gemma-4-26b-a4b-it`.
  • AI Studio > Rate Limit for the same project shows:
    `Project quota tier unavailable. Please contact your project administrator for assistance.`

Why this does not look like a local code/config issue:

  • The same API key can call `models` and `countTokens`, so it does not look like a missing or invalid key.
  • The API is enabled and the project is active.
  • The key is restricted to `generativelanguage.googleapis.com`.
  • The bound service account is enabled and has Service Usage Consumer.
  • We also tried a fresh organization-owned free-tier project. It briefly served one successful request and then converged back to the same project-denied 403 within minutes.

Could someone from the Gemini API team check whether this project is blocked by a hidden generation-serving review, quota-tier mapping issue, or project access restriction? If there is a specific appeal or review path we should use, please point us to it. We can also submit the unexpected-403 form with the project number if that is preferred.

Thanks in advance.

2

Hello @Sion_Jeong ,

This is very likely caused by an issue which is currently only visible in Google Cloud console. We’re working to move visibility of this into AI Studio. To resolve this:

Open the Cloud console & select your project
You should see a banner which explains the issue / may allow you to appeal

3

Hello Pooja,

Thank you for the response.

I checked Google Cloud Console for the affected project and also reviewed the project dashboard, APIs & Services pages, and organization resources, but I am unable to find any restriction banner or appeal option.

The affected project is:

Project ID: petlog-gemini-20260623c

The issue remains:

  • models.list → HTTP 200

  • countTokens → HTTP 200

  • generateContent → HTTP 403

  • “Your project has been denied access. Please contact support.”

Could you please advise where the banner should appear, or confirm whether there is a project or organization-level restriction that requires manual review?

Thank you.

4

Hello @Sion_Jeong ,
If you are unable to see the banner, please contact Cloud Billing Support.