I’m having the same “User location is not supported for the API use” (400, FAILED_PRECONDITION) issue that several others have reported here recently.
Timeline:
I set up billing (prepay) on this project about a week ago.
Everything worked correctly for about a week, including gemini-3.5-flash, gemini-3.6-flash, and explicit context caching (cachedContents).
The API suddenly started returning “User location is not supported for the API use” with no configuration changes on my end.
What I’ve already tried:
Created a brand NEW API key in the same project — same error persists. This suggests the issue is at the project/billing-account level, not the API key itself.
Redeployed my backend server (hosted on Render, Frankfurt region) to potentially get a new outbound IP — same error persists.
Confirmed the server’s public IP resolves to Frankfurt, Germany (not a VPN or restricted region).
Our production voice AI service started getting “User location is not supported for the API use” (WebSocket close code 1007) on all Gemini Live API connections starting around 10:30 UTC on August 17, 2026. Canada is a supported region per Google’s docs.
After investigation, the issue is specifically IPv6. From the same server (OVH datacenter in Canada):
Node.js v20+ defaults to IPv6 via Happy Eyeballs, so all our API calls were going out over IPv6 and getting rejected. Multiple servers with different IPs in the same datacenter are affected. Workaround: --dns-result-order=ipv4first
–no-network-family-autoselection in NODE_OPTIONS.
This was working fine until today. It appears Google’s geolocation database is misclassifying some OVH IPv6 ranges as unsupported regions.
Thanks bro. I had the same issue. Using litellm, the server is in akami’s CA data center. It was indeed caused by ipv6. Let AI adjust to prioritize ipv4, and the problem was solved. Never would I have thought it would be this reason. It should be a problem with google’s ipv6 geographical judgment. Thanks!