Skip to content

Gemini CLI returns 429 RESOURCE_EXHAUSTED on Google AI Pro account after full local reset #22545

Description

@edwardgushchin

What happened?

Gemini CLI consistently fails with 429 RESOURCE_EXHAUSTED / rateLimitExceeded when I use Google login on a paid Google AI Pro account.

This reproduces even after a full local reset:

  • uninstalled and reinstalled Gemini CLI
  • removed ~/.gemini
  • removed project ./.gemini
  • removed C:\ProgramData\gemini-cli
  • removed global npm package files
  • cleaned npm cache

It also reproduces from a completely empty directory.

AI Studio in the browser works with the same Google account, but Gemini CLI fails immediately against the Code Assist backend.

What did you expect to happen?

Gemini CLI should work normally with Google login on a paid Google AI Pro account and should not immediately fail with 429 RESOURCE_EXHAUSTED after a full local reset.

Client information

  • CLI Version: 0.33.1
  • Git Commit: f59f9e9
  • Session ID: a78b9dce-df97-49da-bc12-4753bf082702
  • Operating System: win32 v24.14.0
  • Sandbox Environment: no sandbox
  • Model Version: auto-gemini-3
  • Auth Type: oauth-personal
  • Memory Usage: 211.3 MB
  • Terminal Name: Unknown
  • Terminal Background: #0c0c0c
  • Kitty Keyboard Protocol: Unsupported

Login information

Google login.

Account has an active paid Google AI Pro subscription.

I am not using an API key for this repro.

Anything else we need to know?

This does not look like a local install or workspace issue.

What I verified:

  • reproduced on Gemini CLI 0.33.1 and 0.31.0
  • reproduced in different folders/workspaces
  • reproduced from a completely empty directory: C:\temp\gemini-clean
  • reproduced after full local wipe and reinstall
  • reproduced with different models, including gemini-3.1-pro-preview and gemini-2.5-flash
  • AI Studio web works with the same account
  • GCP / AI Studio project quota pages do not show corresponding exhaustion

The failing backend is:
https://cloudcode-pa.googleapis.com/v1internal:generateContent

I also previously saw the same issue on:
https://cloudcode-pa.googleapis.com/v1internal:streamGenerateContent?alt=sse

Observed trace IDs:

  • 28a721bf01049e41
  • 381891ced4795c2b
  • 33879b0ea6fe71fc

Representative error:
429 Too Many Requests
RESOURCE_EXHAUSTED
rateLimitExceeded

This looks like an account-side Gemini CLI / Code Assist quota or entitlement issue rather than a local machine problem.

Metadata

Metadata

Assignees

No one assigned

    Labels

    area/platformIssues related to Build infra, Release mgmt, Testing, Eval infra, Capacity, Quota mgmtstatus/need-triageIssues that need to be triaged by the triage automation.status/possible-duplicate

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions