The best mini-split air conditioners

Mini-splits are the most efficient way to cool a single room, and the way that efficiency gets quoted makes the gap look larger than it is.

The ENERGY STAR Certified Mini-Split Heat Pumps dataset holds 14,328 US records. We filtered to single-zone residential capacities — 9,000 to 24,000 BTU, which is 9,129 records — and ranked on:

1. EER2 first, because it is measured at a rating condition and is therefore the only figure comparable to a window unit’s CEER.
2. SEER2 second, as the seasonal cooling figure.
3. COP at 5°F for anyone in a climate where the heating half matters.
4. Refrigerant GWP, as a tie-breaker.

We did not rank on noise, because no mini-split publishes a certified noise figure in this dataset. Indoor units are genuinely quiet — that is one of the category’s real advantages — but every decibel number you see is a manufacturer’s own.

Two figures worth holding onto. 12,434 of the 14,328 certified units carry the Most Efficient designation — 86.8% of the field. And 12,971 meet the peak-cooling requirement.

A label that 87% of a category carries is not a shortlisting tool. It is a floor. This is the opposite of how selective the same label is elsewhere — 35% of room air conditioners, 15% of ceiling fans, 9% of bathroom fans. In mini-splits you have to compare the numbers directly.

This is the single most useful thing on the page, and it is where most mini-split marketing quietly cheats.

SEER2 rates a mini-split across a modeled cooling season including a lot of part-load running, where an inverter compressor is at its most efficient. CEER rates a room air conditioner at a single rating condition plus standby. They are different test procedures measuring different things.

Putting SEER2 33.5 next to CEER 17.6 and concluding a mini-split is “almost twice as efficient” compares a seasonal average to a peak-condition rating. It is not a comparison.

EER2 is the mini-split’s rating-condition figure, and it is what belongs beside CEER:

• Best certified mini-split: EER2 19.1 at 9,000 BTU.
• Best certified window unit: CEER 17.6 at 9,500 BTU.
• That is roughly a 9% advantage at comparable conditions.

The seasonal advantage is real and larger than 9%, because an inverter running gently for long periods genuinely beats a unit tested only at full load. But it is not the 90% that the SEER2-versus-CEER framing implies. The full head-to-head.

Every unit in this dataset is a heat pump. That is the category’s real argument, and the cooling numbers are almost a side effect.

• COP at 5°F is the figure that matters for heating in a cold climate, and it is published per model. The units here run 2.20 to 2.48 — meaning 2.2 to 2.5 units of heat per unit of electricity at 5°F.
• Cooling efficiency barely predicts heating. Across the certified field the correlation between SEER2 and COP at 5°F is very weak, so a top SEER2 number tells you almost nothing about winter performance. The full analysis.
• HSPF2 is the seasonal heating figure, and it spreads wider than the cooling figures: 11.80 to 14.00 across these three.
• A window unit with heat exists too, and the best reaches CEER 17.6 — the 33 certified reverse-cycle room units. It will not match a mini-split in genuine cold, but it is a fraction of the installation.

1. Ask for EER2, not just SEER2. Both are published for every certified model. A quote that only mentions SEER2 is quoting the flattering one.
2. Ask for COP at 5°F if you have winters. HSPF2 is seasonal and can hide poor cold-weather performance.
3. Check the refrigerant. R-410A at GWP 2,088 is still 38% of the certified field. R-454B (470) and R-32 (675) are the modern choices.
4. Ignore the Most Efficient badge here. 87% of the field has it, so it does not narrow anything.
5. Get the outdoor unit sited properly. Clearance, shade and drainage all matter, and none of it is on the spec sheet.
6. Price the window alternative honestly before committing. For one room with a usable window, the efficiency gap at rating conditions is about 9%.