Show HN: Sunk Cost – How long until a local LLM rig pays for itself?

Which open models fit your machine, how fast they run, and how many tokens it takes to pay back the hardware against an API. Honest enough to say never.