We are using the Gemini Vertex AI API (Bi-directional API), it was working fine but from the last 2 hours we have been running into these error. We have tried debugging it but from our end, we are sending the data bidirectional API and receiving this error. Looks like something is down related to GO packages.
Hi @legacy-dev ,
Welcome to Google Cloud Community!
The error messages “Unsupported client message: go/ debugproto” and “go/ debugstr” suggest an issue on the server-side (Google’s Vertex AI). These messages indicate that the Gemini API is receiving debugging messages from a Go client (likely your code, though unintentionally) that it doesn’t know how to handle. Since this occurred abruptly after working fine, a recent change or bug on Google’s end is the most likely cause. Here are some potential solutions that might address your questions:
- Check Google Cloud Status Dashboard: First and foremost, check the Google Cloud Status Dashboard (search for “Google Cloud Status” on Google). Look for any reported incidents related to Vertex AI or related services, particularly those involving WebSocket communication or API endpoints. Google may already be aware of the issue and working on a fix.
- Implement Exponential Backoff with Retry Logic: Implement a reliable retry mechanism with exponential backoff. If the API returns this error, wait a short period (e.g., 1 second), then retry the request. Gradually increase the wait time with each subsequent failure. This can help your application recover automatically once the issue on Google’s end is resolved. Make sure to limit the number of retries to avoid infinite loops.
Was this helpful? If so, please accept this answer as “Solution”. If you need additional assistance, reply here within 2 business days and I’ll be happy to help
Hi Ibaui - I implemented the exponential backoff and it did not work at all. Thank you for confirming it is a Google issue - I have a support ticket out since last Thursday and have not heard any updates from the google team. This is preventing us from getting full functionality working for our customers.

