gemini-3.1-pro-preview is hitting intermittent timeouts right now. Root cause is Google upstream capacity pressure. Even with ample account-side concurrency, the upstream still returns status_code=429, Resource has been exhausted (e.g. check quota). — an upstream quota exhaustion at Google that cannot be absorbed on our relay side.
💡 Suggested mitigations (use your own judgment per task):
- Raise the client-side timeout to match task complexity — give the upstream a generous window for long-running calls
- Add client-side retries with exponential backoff — most 429 hits recover on retry
- If your task tolerates a downgrade, you can optionally call
gemini-2.5-proinstead — that model is currently stable
← Back to Live Updates · 📚 Archive