I’m seeing the same, since 6am Eastern today.
Is this the right status page? There has not been an incident report. Google AI Studio
We’re paid tier 1. Same issue.
I have never seen an issue produced for gemini outages, it’s frustrating. I’d recommend filing support tickets to put some pressure on GCP to take these outages seriously. You can go to your projects support @ https://console.cloud.google.com/support/cases
Which product to choose? I am not seeing a non-Vertex-AI Generative AI / Gemini API product listed there.
Its really frustrating when this occurs, and completely puts me off utilising in the case of production applications. I am having to write custom monitoring specifically for the Gemini models so that I can track and report these incidents internally.
I really feel that Google should consider including model overloads as an outage category.
We are running at 5% our assigned Gemini capacity (paid user with agreed raised tiers).
Its Gemini 2.0 Flash having the issue.
Gemini 2.5 Flash is not overloaded still, but we can’t switch to it as it underperforms for our use cases.
Totally agree. We’ve been relying on Gemini 2.0 Flash on some critical tasks and it’s outperforming 2.5 Flash AND 2.5 Pro.
I don’t know what happens to Google, but they haven’t been really showing much care about the developer community. At OpenAI’s Dev forum, there are a bunch of OpenAI engineers and PMs helping and listening.
Google, if you are listening, talk to us about what use cases we care about and train/release models better than previous models not worse. Do NOT overfit to benchmarks!
Hey folks, sorry for the issues here. The 2.0 Flash issue should be resolved, please let me know if not.
When can we expect the issue to be solved for larger, more recent models like 2.5 Pro? Right now, if I try to parse a document with >8 pages I have to implement retries with some exponential backoff, with delays hitting more than 1h since the server keeps disconnecting. This makes building production apps with Google Gemini impossible 
1 Like
Hey Vishal
I am still facing the issue with Gemini-2.0-flash. I am using retries and backoff with 5 and 5 values.
503 UNAVAILABLE. {'error': {'code': 503, 'message': 'The model is overloaded. Please try again later.', 'status': 'UNAVAILABLE'}}
Could you please tell, by when will this be resolved
Thanks a lot
Hello folks!
Is someone experiencing error 503 with the Gemini API these past days and also today? I went to the status webpage (Google AI Studio), where it says the API have been working well the last days.
By the way, the model I’m facing more problems with is Gemini 2.0 Flash.
Yeah I’m seeing the same, today in particular…basically not working at all! I’ve been testing for bugs on my side, but can’t see anything besides the 503 and the service being overloaded/unavailable…anyone else?
Same for me. Swapping from gemini20Flash to gemini20FlashLite solved the problem (for now).
Seriously considering switching to OpenAI.
