My project is set up with pay as you go billing model. But when trying the model gemini-exp-1114 I have a very low limit - like 2 or 3 per minute and I hit RESOURCE_EXHAUSTED very early. Does this model support higher pay as you go limits ?
1 Like
Welcome to the forums!
The exp-1114 model is, as the name suggests, experimental only and not billed for access. Since it is not billed, it is subject to the lower limits.
1 Like
is the limit same as free tier then? Where can I look at the token limits for experimental models ?
1 Like
Based on testing, the usage limits for experimental 1114 are the same as free tier Gemini pro. The input token limit you basically see in AI Studio, mine shows 32k.
Hope that helps.
1 Like
Is that 32k per day?
1 Like
No, it’s per request.
The limit for the experimental model might be around 2-3 rpm (requests per minute), if I’m not mistaken.
1 Like
I am below 2 rpm but I get resource exhausted after 3 -4 requests. Are there other limits? Like number of tokens per minute or something.
Sorry, I don’t know. Maybe a Googler can elaborate on that.
I’m currently exploring the OpenAI compatible endpoints.