[lorey]

Doesn't this depend a lot on private vs company usage? There's no way I could spend more than a few hundreds alone, but when you run prompts on 1M entities in some corporate use case, this will incur costs, no matter how cheap the model usage.