The limit varies based on demand, now with the release of the new model it's very limited, but it wouldn't be like this forever.

Honestly, like I did with GPT-4 and GPT-4-turbo, I would just use the API on an alternative frontend.