Feature request: rate limits control #204
Description
For users to actually put this to production (a fancy word for deploying to the web, usually) we must enforce strict rate limits so we don't wake up to an outrageous bill from our beloved LLM provider just because someone really wanted to experiment with our little app.
Following a discussion on LinkedIn I understand this already exists in the underlying packages, so bringing this as an essential argument/configuration step is a must.
Another idea that came to mind is to prompt users to provide their own api keys, and in that case, some mechanism of multiple models should be implemented (in ellmer?).
Regardless of how you tackle this idea, please don't just encourage people to naively share their api key with the world, which is exactly what's happening when they provide this free & limitless access to query the model with their key in the background...