Which power station should I buy?
Each list below answers one buying question by running the same set of published station records against one load, or one set of loads, and ordering what survives. The rule is four gates — does it start and hold the load, can we prove how long it lasts, can it be bought today, and only then price — and every list prints that rule above its own table.
They are grouped here by what the arithmetic actually returned, because that is the part you cannot see from inside any one of them: the loads where the choice of machine decides the outcome, the loads every machine handles so the real question is duration, and the loads nothing on this site runs at all.
Every list here ranks the same 12 US records — one mains standard, one currency, one set of prices. The library holds 26 station records with a page: the other 14 are EU versions of the same machines, ranked on the load pages rather than here — a machine built for a different mains standard is not a cheaper alternative, it is a different product at a different price. That is the difference between the two counts a reader meets on this site: a load page ranks the library, a list here ranks one region of it.
Some machines run it, and which ones is the whole question 6 lists
Hardest first — the fewer machines clear a load, the more the choice of machine decides the outcome. Every refusal on these lists is a named constraint (continuous output, a start-up ceiling, an inverter waveform), not a shortlist someone curated.
Every machine we publish runs it 4 lists
Here the buying question is not “which one” but “for how long”, so these lists are ordered by the runtime we can prove rather than by the verdict. Clearing a load and being worth buying for it are different claims, and the second is the one these tables settle.
No machine here runs it 2 lists
These lists recommend no machine and sell nothing. They exist because the answer is genuinely no on today’s records, and the page names the constraint that stops each one — which is the part that tells you whether a bigger battery would have helped.
How far the runtime figures reach
5 of the 12 lists print a proven runtime for their pick. On the rest the verdict stands but the duration does not: either no duty cycle is published for the load to divide against, or no usable-capacity figure is published for the model, and dividing a nameplate by a label watt figure overstates the answer by a fifth or more. Each list names which of the two stopped it.