Local LLM Speed by GPU Generation, and Our Production's Daily Token Volume

Checked with real logs instead of gut feeling In my last post(new tab) I talked about switching to...