> OpenRouter has Gemma4 31b at ~38-50 cents per million tokens. This means that on the optimistic side (50 watts, 40 tokens per second, and 10 years) the pro max is as cheap as openrouter. On the pessimistic side (100 watts and 3 years at 10 tokens per second) the pro max is 10x the cost. I think ~3x the cost per million tokens is likely the right number for local inference on the pro max from an accounting perspective.

Apart from that, like detailed in the the article, pricing for local compute also depends on electricity prices.

By the way, I don't want to snark about it, my English is not very good, but it's "per se", not "per say". Just commenting on this petty thing because it seems to be a common misspelling, and it always trips me up a bit. Makes me wonder about another supposed meaning like "from hearsay".