Continuously experiencing 429 and 503 errors in Gemini Models

I am continuously experiencing 429 and 503 errors with Gemini 3.8 Flash/3.7/3.6. Even with retries and proper backoff, requests frequently fail again, causing incomplete outputs, increased processing time, and unreliable production workloads.

Please investigate the recurring 429 rate-limit and 503 service-unavailable errors, including service capacity, quota behavior, and recommended retry settings. Improved API reliability for batch processing is urgently needed.

Hello @indranil_kundu, if you are receiving 429 errors and have confirmed on the Rate Limit Dashboard that no rate limit was breached during that period, please submit your details using this form so we can look into this for you.