Comment by barrystaes

Fun, but this seems to assume all LLM run on SAAS subscriptions. I would like to compare this to local LLM costs by converting my usage load x hardware costs into a token price. In addition if it games the LLM so often, these eventually are optimized and basically cheat on the score.