Frequent 503 Errors (Service Unavailable) across all models

Gemini API
,
108

Error 503, Message: This model is currently experiencing high demand. Spikes in demand are usually temporary. Please try again later., Status: UNAVAILABLE, Details: []

Issue is not resolved, I’ve been getting 503 errors in almost every API call throughout the whole day.

109

Unbelievable, I face this for 2 days, and I’m also a paid user.

110

I am having this issue since this afternoon. All models

111

Just tried using other providers inside genai SDK for Go and the 429/503 errors persists.
Meanwhile, the official page ( Google Cloud Service Health ) reports no problems at all when users are reporting 40% error rate in their API calls. That is just unacceptable.

112

Gemini ultra user here, problem not fixed and it actually consumed credits as i was trying.

wasted time and money on this subscription

113

This is still on going. Over the past 3 days my app has had over 8000 500 errors, it’s entirely unusable. Is Google working to address this?

114

i am getting this same error last two weeks. posted couple of times. but no reply from any expert of support. so much unhappy with this.

“Model busy, timeout or network error, retrying in 5895ms…”
Error
“Error processing image:” “{"error":{"code":503,"message":"This model is currently experiencing high demand. Spikes in demand are usually temporary. Please try again later.","status":"UNAVAILABLE"}}”
Log
“Model busy, timeout or network error, retrying in 10303ms…”

115

Agreed @Chris_Mullins , have faced > 2000 503 errors on even older models like 2.5 flash.

@Logan_Kilpatrick - can you atleast share some update as even paid users are impacted…
how can we enable applications for people in this form ?

116

Unbelievable…

{
  "error": {
    "code": 503,
    "details": [
      {
        "@type": "type.googleapis.com/google.rpc.ErrorInfo",
        "domain": "cloudcode-pa.googleapis.com",
        "metadata": {
          "model": "gemini-3-flash-agent"
        },
        "reason": "MODEL_CAPACITY_EXHAUSTED"
      }
    ],
    "message": "No capacity available for model gemini-3-flash-agent on the server",
    "status": "UNAVAILABLE"
  }
}
117

It might be beneficial to establish a group (perhaps on a platform like WhatsApp) to collectively ascertain if our individual applications are indeed encountering an identical issue. This would obviate the need for each of us to investigate personal causes.

Furthermore, it might be prudent to forge a professional advocacy group to establish a direct conduit for dialogue with Google ?

118

It’s been days since I’ve been able to get an api call through to Gemini 3.1 Flash Lite. Gemma4 works fine consistently, though, it just has limits.

119

Same issue here as a paid user. It’s almost unsuable..

120

Same here, I’ve been reporting this over different posts with not even a single explanation or follow up.

I just did a query to check and we did more than > 3000 503/429 errors over the last few days. OpenRouter says the the API Uptime is below 90% in 2.5 models (flash and pro), but the official API status page does not reflect that.

121

Hi there. Same issue here. I’ve facing the 503 errors for more than a mounth. It is frustanting. I have changed from google ai studio for vertex ai and there is no difference. I’m developing a product that uses gemini 3.1 pro preview as api and its is impossible to teste and put it to production sice almost every day a have a 90% rate failure with 503 errors. I tought it would be solved in about a mounth, but it seens it doesn’t. How longo do we have to wait for a stable version that could work for our products? it will be solved one day? I would not like to change for another model.

122

This is still an issue, even when using the paid subscription. Very frustrating.

123

in my experience,timing is key factor.usually before utc 09:00, it is available.

124

same here… working with n8n self-hosted, call the gemini api and yes, most of the time the workflow stop due to 503 error, so frustrating

125

it’s unusable. leaving today.

126

I am shocked, how is possible to be on ultra plan paying almost 600 $ for a product that i have to wait response for 10-15 minutes, opening new terminals without result, but that’s another topic.
I had same 503 issue combined with too many requests error (sending 5 requests not 500) even in the contract they say 1000 tpm , now I am on tier 1 looking forward to spend 10$ so i can pay and leave this forever.

127

Hello Everyone,
so sorry you’re still experiencing this. We have some guidance on using retries in this pinned post on 503’s / 429’s