Persistent 500 INTERNAL Server Errors from Gemini API (generate_content) - Requesting Assistance

Gemini API
1

Hello Gemini Developer Support Team,

I am writing to report a recurring issue with the Gemini API’s generate_content endpoint that is currently blocking a production workflow on my end.

Issue Summary:
When calling client.models.generate_content() through the google-genai Python SDK, I am consistently receiving a 500 INTERNAL server error. The request fails even after automatic retries (3 attempts with exponential backoff) and after falling back to a second API key, both of which ultimately fail with the same error.

Error Details:
google.genai.errors.ServerError: 500 INTERNAL
{
“error”: {
“code”: 500,
“message”: “An internal error has occurred. Please retry or report in https://developers.generativeai.google/guide/troubleshooting”,
“status”: “INTERNAL”
}
}

Environment:

  • SDK: google-genai (Python)
  • Python version: 3.11
  • Retry library: tenacity
  • Both configured API keys returned the same server_unavailable / 500 INTERNAL error

Steps to Reproduce:

  1. Load a batch of ASINs/items for processing
  2. Call generate_content() with prompt_kind=“life_info”
  3. Observe repeated 500 INTERNAL errors across multiple retry attempts and across multiple API keys

Impact:
This is affecting a real-time data processing pipeline, and the repeated failures across two separate API keys suggest the issue may be on the Gemini API’s server side rather than with my request payload or account configuration.

Could you please help investigate this issue and let me know:

  1. Whether this is a known/ongoing service issue
  2. If there is anything on my end (request format, quota, region, etc.) that could be causing this
  3. Any recommended workaround while the issue is resolved

I am happy to share additional logs, request payloads, or timestamps if that would help with debugging.

Thank you for your time and support.

Best regards,
Nitin Pawar

2

since its happening across two separate keys, that does point away from an account or quota issue. worth checking if its tied to a specific payload size or a field in your prompt_kind data, big batches or odd characters can sometimes trip a 500 on their end. also check https://status.cloud.google.com for any ongoing incidents, and the troubleshooting guide here: https://ai.google.dev/gemini-api/docs/troubleshooting