429 RESOURCE_EXHAUSTED on Tier 1 with available quota

Hello all!

I’m getting 429 RESOURCE_EXHAUSTED on every API call to gemini-2.5-pro, even a single request.

Setup:

  • Project: Google AI Studio, Tier 1 (billing enabled, confirmed in dashboard)

  • SDK: google-genai 1.56+

  • Rate Limit dashboard shows: 637/1000 RPD, 30/150 RPM — quota is NOT exhausted

  • Server: Railway (US region)

Error:

429 RESOURCE_EXHAUSTED {'error': {'code': 429, 'message': 'Resource has been exhausted (e.g. check quota).', 'status': 'RESOURCE_EXHAUSTED'}}

What I tried:

  • Created a new API key in the same project — same error

  • Migrated from deprecated google-generativeai to google-genai — same error

  • Earlier today, a different key returned free_tier_requests violation. After switching to correct Tier 1 key, error changed to generic RESOURCE_EXHAUSTED

Question: Is there a hidden cooldown or project-level throttle beyond what the Rate Limit dashboard shows? Has anyone seen 429 on Tier 1 with available quota?

Same problem here. They have a bug. They changed something since 3-4 days ago. Actually after 3.1 Pro was released, they broke this.

I have the same problem but for Batch API although im Tier 3, without a single running batch job.

This comment of mine might help: 429 RESOURCE_EXHAUSTED - blocks legitimate platform usage - #7 by abbm

In short, Google’s API returns 429 when their servers are overloaded, not just when you’ve exhausted your quota!