Can't reach any gemini 2.5-flash models . All of them return 500 error

Gemini API
, ,
1

Hello everyone,

First of all, I’d like to say that this is my first post, and I hope I’m not breaking any rules. I’ve tried searching for a solution to my problem but couldn’t find anything similar, so here it goes. Hopefully, someone can lend me a hand.

I have an app where (my mistake) I was calling the gemini-2.5-flash-preview-05-20 model, using its free tier (I don’t have a payment method added) from Google AI Studio.

This model was deprecated a few days ago, so when I realized it, I started looking for a free tier model that could replace it, but the documentation is a bit confusing. I’m seeing different model names on some pages compared to others.

The thing is, I’ve tried using the following names (endpoints):

  • gemini-2.5-flash-preview-04-17
  • gemini-2.5-flash-lite-preview-06-17
  • gemini-2.5-flash

And I’m sure I’m forgetting a few others. All of them return the same 500 error. I don’t know what else to try because I can’t find any other names, and I find it very strange that there isn’t another 2.5-flash endpoint available. Could it be that there’s no longer a free tier for this model, even though the documentation says there is?

I add more logs to my app and that’s what I can see:

message: “[GoogleGenerativeAI Error]: Error fetching from https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash:generateContent: [500] An internal error has occurred. Please retry or report in Troubleshooting guide  |  Gemini API  |  Google AI for Developers”

I also find it strange that no one else seems to be running into this problem, so it must be something on my end. That said, if I use the 2.0-flash model, it does work.

Regards,
Santi.

3

Hey @barallat,

Welcome to the Forum!

It sounds like the 500 errors might be due to deprecated or unsupported model endpoints. If gemini-2_0-flash works but the 2.5 versions don’t, it’s possible the free tier access for those newer models is limited or temporarily down. You might want to stick with 2.0 for now or check with support for updated model availability.

Thanks!

4

Hey @Deepakishore thanks for your advise!

is what I have now but sounds weird to me… Maybe can we have a word from Google??

5

Hi @barallat,

Please wait and try your request again as a 500 error usually indicates an unexpected error on Google’s side.

Thank you!

7

I’ve taken another small step forward in debugging this issue. I’ve now isolated the trigger for the 500 error, but I’m not sure about the root cause.

In my request, I’m sending a video to Gemini for analysis. If I remove the video from the payload, the request to the 2.5-flash model completes successfully. However, with the 2.0-flash model, the request including the video works perfectly fine.

The video file is small, between 1 and 5 MB. I’ve tested with both .mp4 and .webm formats. I have also tried using API keys from projects with no billing enabled, as well as keys from high-billing projects… It makes no difference; I always get a 500 error when a video is included in the request to the 2.5 models.

Can anyone confirm if the gemini-2.5-flash and gemini-2.5-pro models have specific limitations or bugs related to video input?

Thanks for any insights.

1 Like