Me too, exactly this.
Hard to trust a system so janky, when we are all writing code that relies on models working
This is google we are talking about
For me as well, Gemini Is not working
this feels sucks, My api completion rate is at 20%
Been getting the 503 error for the past 2 days. An update on when this might be resolved would be wonderful.
Esse mesmo problema venho enfrentando a mais de uma semana. Tenho o plano pago e a API com faturamento configurado e ativo. Já migrei para o novo modelo, mas mesmo assim não funciona! Preciso de uma solução urgente!
We’ve been hearing that the team is “aware” of the issue since January. We are now in March, and these 503 errors are still happening at an incredibly high frequency.
This level of instability is a massive blocker for developers.It’s nearly impossible to rely on the API right now.
Could you share some concrete details on the team’s actual latest progress? “Shuffling capacity” doesn’t give us much to plan around. Are there any solid timelines or structural fixes rolling out soon that we can look forward to?
Frequent 503 errors are occurring and severely impacting my service. Please optimize the system as soon as possible so I can have peace of mind. Additionally, I would like a refund for the money that has been wasted due to this issue.
Where did you hear the team is aware of it?
You can check the official team’s reply in this thread from January 26th. They marked the post as resolved when they replied, but the issue wasn’t actually fixed, and they haven’t responded to anyone since.
The issue is definitely not resolved. I am getting errors on all Gemini 3 models for days, unable to use GEMINI API at all for at least 3- 4 days with Gemini 3 Pro, Gemini 3 Flash or Gemini 3.1 Pro preview…
Just contributing to feedback here… Issue is not resolved. Quota exceeded messages prevent further use.
hope is they are trying to resolve it and that’s why it’s down for everyone…
Not resolved by a long shot. Enough to look at downdetector to see the cyclic up/down across 24hrs every day. But even without downdetector, just invoking the generate API you can see immediately it doesn’t work half+ of the time.
@Logan_Kilpatrick FYI.
This has not been resolved. I have been getting the same error since the past 3 days and I absolutely cannot go ahead and test and release without this getting solved. Seriously considering shifting away from Gemini. I am on the paid tier as well.
@Logan_Kilpatrick Please look into this, the issue is far from resolved.
Facing this issue since last night across all models in Antigravity. Suddenly the other models quota is showing 0% with i week of lock out an gemini flash is constanly giveing
UNAVAILABLE (code 503): No capacity available for model gemini-3-flash on the server
I am seriously considering shifting to something else
I face this for 3 days, and I’m paid user, this is incredible!
@Logan_Kilpatrick I have an AI app builder called episolo, which uses only Gemini models and many users left because of this issue. Getting 503 for nearly all requests!
Even with a paid Tier 3, we get regular timeouts and 503 on gemini-3-flash-preview. We are literally waiting for GA models on 3 series… this is so exhausting to work with. None of the 3 series are GA and we are finding it really difficult to use in production.
HAPPENS OFTEN: [GoogleGenerativeAI Error]: Error fetching from https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash:generateContent: [503 Service Unavailable] This model is currently experiencing high demand. Spikes in demand are usually temporary. Please try again later.