Persistent 403 Forbidden Error Accessing Google AI APIs from VPS in Düsseldorf, Germany - Affects Multiple Endpoints & Models

Gemini API
, ,
1

Dear Google AI Support Team,

I am writing to report a persistent 403 Forbidden error I am encountering when attempting to access Google AI APIs from my Virtual Private Server (VPS) located in “Düsseldorf”, Germany.

Key Issue:
All API requests to Google AI services from this specific German VPS result in a 403 Forbidden error with the message: “Your client does not have permission to get URL [URL_ENDPOINT] from this server.”

Important Context & Troubleshooting Done:

  1. Successful Access from Other Location: I have tested the exact same API keys, request bodies, and scripts from my friend’s VPS , located in Singapore , and all requests complete successfully without any 403 errors. This strongly suggests the issue is specific to the German VPS environment or its network path to Google’s services, rather than a general problem with my API key configuration or the API services themselves.

  2. Affects Multiple Endpoints: The 403 error occurs for both:

  • Native Google AI Studio Endpoints: e.g., (and other v1beta model endpoints like gemini-2.5-flash-preview-04-17).
  • OpenAI-Compatible Endpoints: e.g..
  1. Affects Multiple Models: The issue is model-agnostic. I have experienced this with various models, including gemini-1.0-pro, gemini-2.5-flash-preview-04-17, and gemini-2.5-pro-exp-03-25.

Here is an example cURL command for the OpenAI-compatible endpoint that consistently fails from my German VPS:
curl -X POST “urlv1betaopenaichatcompletions”
-H “Content-Type: application/json”
-H “Authorization: Bearer AIzaSyDz-W5G3e9_TXtimbrfaxxXxxxxxxxx”
–data-raw ‘{
“messages”: [
{
“role”: “user”,
“content”: “what model are u. Are u from Google? Perkwenalan dikit lah detail model mu”
}
],
“model”: “gemini-2.5-pro-exp-03-25”,
“max_tokens”: 508,
“temperature”: 0.6,
“top_p”: 0.9,
“stream”: true
}’

Response Received from German VPS:

Error 403 (Forbidden)!!1

403. That’s an error.

Your client does not have permission to get URL /v1beta/openai/chat/completions from this server. That’s all we know.

(Please note: A similar 403 error was also received for the native endpoint /v1beta/models/gemini-2.5-flash-preview-04-17:generateContent using API Key starting with AIzaSyDfFn4… from the same German VPS, as discussed previously.)

This issue is significantly hindering my ability to develop and deploy my AI application that relies on your services for users potentially in or routed through this region.

Could you please investigate if there are any known issues, specific restrictions, or network-related problems for accessing Google AI APIs from the Düsseldorf, Germany region, or perhaps from the network of my VPS provider rarecloud io Any insights or guidance you can provide would be greatly appreciated.

Thank you for your time and assistance.

3

Same issue here, my VPS is located in Frankfurt, DE. I tried running the same curl command on both the VPS and my local machine with the same api key and my local machine (Canada) got a response with no issues at all. Hope this gets fixed