They basically saying they cant solve it till management allocates more servers. And this been going since last year so looks like we should move to another provider
Hello, I hope I am in the correct place. I tried to get help from Google Cloud Support but they told me to come here. Let me first say I am not a developer. I am a chiropractor that used to be a computer geek. I used Google Ai Studio to help build an app for my business and I published it just for use within my business. Over the last couple of months things have worked great but as of sometime between last night and this morning I have lost my database of patient files within the app. All that is left is a few dummy files that were created when I was making the app before publishing. It is like it reverted back to the start. The app itself is fine but all my data is gone. I have not received any errors on my end. How can I get this data back?
I am getting a 429 error despite not having used that project(it has just 1 api key) in 2 days. the graphs despite showing i have not used any model in past 2 days, the table shows that i have maxed out my RPD 27/20 in gemini 2.5 flash i am considering having to make a new project altogether 
Same here so far flash 2.5
Hi,
I’m reporting recurring 503 “model is overloaded / high demand” errors on the Gemini API (generativelanguage.googleapis.com) over the past several days, affecting a production chatbot that serves a municipal government website.
Details:
Model: gemini-3.5-flash (streaming, streamGenerateContent)
Error: HTTP 503, status UNAVAILABLE
Frequency: intermittent, several times per day, clustered rather than constant
Not a quota issue: these are 503 (server-side), not 429. My key is well within rate limits.
Key type: created in Google AI Studio (restricted to Generative Language API)
I have also seen intermittent 404 responses on the same gemini-3.5-flash alias, which resolve on retry — suggesting one of the backing endpoints behind the alias may be unhealthy.
I understand 503s are transient and affect all users during capacity spikes, and I’ve implemented exponential backoff plus automatic fallback to gemini-3.1-flash-lite on my side. I’m reporting this mainly so the team is aware of the frequency, since this is a public-facing government service where reliability matters.
Could you confirm whether there’s an ongoing capacity issue with gemini-3.5-flash, and whether the intermittent 404 on the alias is a known routing problem?
Thanks.
Experiencing 503 issues since 4 days with 3.5 flash. At this point this service is completely unreliable for me. I retried many times at different times and I still get the issue. A robust fix would be appreciated. Also, the status page is showing all statuses to green but it’s not the case. I paid credits for a service I can’t even use!