tokencost

Why we included this project

LLM API costs tend to sneak up on you: the numbers look small per request, but they compound fast when an app or agent is calling models all day. Tokencost puts a maintained price table in front of you, covering hundreds of models from the major providers, plus simple functions that estimate prompt and completion costs before you send anything. That makes it handy for teams comparing models, forecasting spend, or deciding which provider should handle a given query. It also counts prompt tokens for OpenAI-style requests, so you get a reliable estimate without building out a full observability stack. It is a small library that does one job well, and the kept-up price data spares you the chore of checking provider pricing pages yourself.