Gemini APi Key verfication not working on models page in Cursor settings

{
“error”: {
“code”: 404,
“message”: “models/gemini-1.0-pro is not found for API version v1beta, or is not supported for generateContent. Call ListModels to see the list of available models and their supported methods.”,
“status”: “NOT_FOUND”
}
}

I guess Cursor is using ‘models/gemini-1.0-pro’ for verifying the api key. I think they have discontinued the use of this model. Even when my api key is valid, it’s not connecting, hope they look into the issue.

6 Likes

I am having the same issue actually. when I change the test to 1.5 pro it works but obviously that’s only for curl, can’t change what model cursor automatically uses

1 Like

I’m having the same problem here. Are you still able to use the Gemini models via API key even though the “verification” fails?


I think the reason bc the outdated model that cursor request. I switch the model to 2.0 in crul, it success.

1 Like

How do you change it in crul?

Just manually do the curl request that it provides you but switch it to the model that Google AI Studio recommends for verification.

However, under the hood Cursor is using a depricated model. I don’t understand why it needs to do this verification step at all. If I put in an API key that doesn’t work with a model I choose, return an error then.

Got the same problem - no gemini models now working through Google API:(

Does following help?

This works for OpenAI as it lets you input an override URL for accessing the endpoint, but not for Gemini.

The verification process for Gemini is just confirming that a request can be sent, but Cursor doesn’t let you override the endpoint.

They should add a way to change the URL for verification, as models can be depricated with short notice leaving a hardcoded verification URL useless.