I’m getting sporadic 429 RESOURCE_EXHAUSTED errors on gemini-2.5-flash-image image generation, despite my usage being well below my Paid Tier 1
limits.
Account Details
- Tier: Paid Tier 1
- Project: Default Gemini Project
- Model:
gemini-2.5-flash-image - Billing: Active, with valid payment method
Usage vs Limits (last 7 days)
| Metric | Usage | Limit |
|---|---|---|
| RPM | 9 | 500 |
| TPM | 1.3K | 500K |
| RPD | 716 | 2,000 |
As you can see, I’m at ~1.8% of my RPM limit and ~36% of my RPD limit — nowhere near the quota ceiling.
Error Details
ApiError: {“error”:{“code”:429,“message”:“Resource has been exhausted (e.g. check quota).”,“status”:“RESOURCE_EXHAUSTED”}}
- Endpoint:
generateContentwithresponseModalities: ['IMAGE'] - Frequency: ~81 occurrences over the past 2.5 months (ongoing)
- Pattern: Appears random — not correlated with high traffic bursts
- Other models unaffected:
gemini-2.5-flash(text) andgemini-2.5-flash-native-audiowork fine on the same API key
What I’ve Verified
- Billing is active and explicitly assigned to the project
- Usage dashboard confirms I’m well within all limits (RPM, TPM, RPD)
- Text-based models on the same key/project work without issues
- The error is specific to image generation requests
Impact
This is a production children’s education app. When users try to generate images, they randomly get failures even though our volume is low. We’ve had to
implement retry logic as a workaround, but the underlying issue seems to be on Google’s side.
Question
Is there a known issue with image generation models being incorrectly routed to free tier quotas on Paid Tier 1 accounts? I’ve seen similar reports from
other developers ([thread 1](https://discuss.ai.google.dev/t/bug-paid-tier-1-account-getting-free-tier-requests-limit-0-on-image-generation-models-gemini-
2-5-flash-image-gemini-3-pro-image-preview/123906), thread
2).
Any guidance on how to resolve this would be appreciated. Happy to provide trace IDs or additional logs if needed.