Hi everyone,
I’m encountering a frustrating quota issue with the Generative Language API using the gemini-2.5-pro-preview-03-25 model.
My project has pay-as-you-go billing enabled and active (confirmed via billing console, incurring costs, valid payment method, Tier 1 usage plan). According to the documentation and my project’s Quotas page, the applicable limit for this model under Tier 1 should be 1,000 Requests Per Day (RPD) .
The Problem: Despite being on the paid Tier 1 plan, my application’s API calls start failing with quota errors after exactly 25 requests in a UTC day. Monitoring confirms we are hitting the specific quota named "Request limit per model per day for a project in the free tier ", which has a value of 25 for this model on my project.
Expected Behavior: The system should allow the first 25 requests for free, and then seamlessly transition to using the paid 1,000 RPD limit without blocking the service.
Actual Behavior: The 25 RPD free tier limit is acting as a hard block, preventing any further requests for the rest of the day and disrupting our application’s core functionality.
Troubleshooting Confirmed:
-
Billing account is active, healthy, and correctly linked to the project.
-
Payment method is valid.
-
Project Quotas page explicitly shows both the 1,000 RPD paid limit and the 25 RPD “in the free tier” limit for gemini-2.5-pro-preview.
-
Attempting to edit the 25 RPD “in the free tier” quota via the console fails, as the UI limits input to a maximum of 25.
-
The API call structure uses the correct model ID and API key authentication (/v1beta/models/gemini-2.5-pro-preview-03-25:generateContent?key=…).
Question:
Has anyone else experienced this specific free tier limit blocking paid usage for Gemini models? Is there a known configuration issue or a step required to ensure the seamless rollover from the free allowance to the paid Tier 1 RPD limit? How can this hard block at 25 RPD be resolved?
Any insights would be greatly appreciated!