What a mini-split costs to run

Every figure in the Watts and Season columns below is computed, not published. The method is one division and one multiplication: capacity ÷ EER2 gives the full-load draw, and that draw over 960 hours at 18.34¢/kWh gives the season cost. Nothing else is assumed.

The spread inside a single capacity is the part worth sitting with. At 9,000 BTU the best and worst certified units are 471 W and 857 W — $68 a season apart for identical cooling output. That is a larger gap than most buyers expect, and it is invisible if you only read SEER2.

Window air conditioners publish a certified annual kWh per model. Mini-splits do not. The certified mini-split dataset holds 18,036 units and carries SEER2, EER2, HSPF2, capacity, refrigerant and COP — and no annual energy field of any kind. There is also no sound field, which is why noise claims for this category always trace back to the manufacturer rather than the program.

So the running cost has to be built from a ratio, and the two available ratios answer different questions:

• EER2 is a single-condition, full-load rating. Capacity divided by EER2 is a real wattage, which is why it is the one you can do arithmetic with. It describes the unit working as hard as it ever works.
• SEER2 is a seasonal average across a range of outdoor temperatures and part-load conditions. It is the better description of a whole summer — and it cannot be converted into kWh without the bin hours and load assumptions buried in the test procedure, which the published dataset does not include.
• The gap between them is the inverter. Across the dataset EER2 averages 12.5 while SEER2 averages 21.4. That spread exists because these units spend almost all their time modulating below full output, where they are much more efficient than the full-load rating suggests.

Which leads to the one rule that matters here: the season figures in the table are ceilings. A real inverter mini-split, cycling and modulating through a real summer, will land under them — often well under. Use them to compare two units against each other, which is exactly what they are good for. Where EER2 and SEER2 sit among the other figures.

The ratings on this page are not typical, and it is worth being precise about how atypical. Of all 18,036 certified units:

• 1,219 reach EER2 15 — 6.8% of the category.
• 435 reach EER2 16 — 2.4%.
• 136 reach EER2 17 — 0.75%.
• 84 reach EER2 18 — 0.47%.
• 20 reach EER2 19 — 0.11%, and the Fujitsu above is one of them.

The whole dataset runs from EER2 7.3 to 21.45. So the practical difference between a carefully chosen unit and a default one is close to a factor of two in full-load draw, and almost none of it shows up in the marketing, which leads with SEER2.

Against the rest of home cooling, at the hours in the table:

• A ceiling fan on low: 2.3 W. 40¢ for the entire season. It cools people rather than air, which is the only reason that number is possible.
• An evaporative cooler: 250 W. $44 a season, and only in dry air.
• A 1,300 W portable air conditioner: $229 a season at nameplate draw — against $83 for the 9,000 BTU mini-split’s full-load ceiling. The comparison in full.

The full league table puts every product on this site in one ranking by watts.

1. Size it to the room, not to the weather. Capacity is the dominant term, and through the small sizes oversizing costs twice — more watts, and a capacity class whose best certified EER2 is lower than the one you left. What size the room actually needs.
2. Read EER2, not just SEER2. Within a capacity, EER2 is the figure that separates 471 W from 857 W. SEER2 compresses that difference and the marketing leads with SEER2.
3. Let it modulate. The whole efficiency advantage of the category comes from running continuously at low output rather than cycling hard. Setting back aggressively overnight and recovering in the morning fights the thing you paid for.
4. Reduce the load before the capacity. Shading and sealing lower the size you need, which is the lever with the largest effect on the table above.
5. Run a fan alongside it. At 2.3 W, air movement buys a degree or two of set point for a rounding error, and set point is run time. Where to set it.
6. Check your tariff before your unit. On time-of-use pricing, when you run it can move the bill as much as which one you bought — which hours matter.