The picks
The best mini-split brands, tested against the certified data
There are 156 mini-split brand names in the certified data and far fewer actual machines. Here is what the badge predicts, and what it does not.
By Sawyer V. · Published
The short answer
Brand barely moves the median. Across the 53 brands with 100 or more certified records, median EER2 runs from 13.0 down to 11.85 — a spread of under 10% for the typical unit. What varies enormously is peak capability (EER2 up to 21.45, COP at 5°F up to 3.28) and cold-climate share (54% to 99%). And one 18,000 BTU model appears under ten different badges.
Fan & Frost earns a commission when you buy through our links, at no cost to you. It never changes a ranking, and we say so when the right answer is the cheaper unit or nothing at all. Full disclosure · How we pick.
At a glance
The picks, ranked
Tap any row to jump to the full reasoning. Prices are live from Amazon where we have them, and read “check price” where we don’t — never a number we cannot verify today.
| # | Product | Best for | Price |
|---|---|---|---|
| 1 | Best peak capability Mitsubishi Electric Mini-Split Heat Pump, 9,000 BTU Mitsubishi Electric holds the highest certified EER2 (21.45) and the highest COP at 5 degrees F (3.28) of any brand in the dataset. | One room where you want the top of what the category can do | |
| 2 | Best median performer Fujitsu Mini-Split Heat Pump, 9,000 BTU Fujitsu has one of the highest median EER2 figures among brands with 100+ records, and this unit is EER2 19.1 with COP 2.48 at 5 degrees F. | Buying a brand rather than researching a model number | |
| 3 | The ten-badge model Carrier Mini-Split Heat Pump, 18,000 BTU 18,000 BTU at EER2 15.3 and COP 2.89 at 5 degrees F — and the identical model number appears under at least ten brand names. | A larger room, bought on price rather than badge | |
| — | Where the brand record disagrees with the model Hisense Mini-Split Heat Pump, 9,000 BTU This unit is EER2 18.35 and cold-climate certified — while Hisense as a brand has one of the lowest median EER2 figures in the field. | A cooling-led installation where the specific model is strong |
We did this analysis for window air conditioners and found 58% of the field sharing a chassis. Mini-splits are worse.
Method
What we ranked on
The ENERGY STAR Certified Mini-Split Heat Pumps dataset holds 14,328 US records across 156 distinct outdoor-unit brand names. We grouped by brand and ranked on:
- Median EER2 — what you get if you buy the badge without researching the model. EER2 is the rating-condition figure, and the only one comparable to a window unit’s CEER.
- Peak EER2 and peak COP at 5°F — what the brand is capable of at its best.
- Cold-climate share — what proportion of the brand’s range is certified for genuine winter heating.
- Most Efficient share, as a tie-breaker — with a heavy caveat, since 86.8% of the whole field carries it.
We did not rank on price, warranty or reliability: none is in the dataset, and we hold no failure-rate data.
The table
The largest brands, by certified record count
| Brand | Records | Median EER2 | Best EER2 | Cold climate |
|---|---|---|---|---|
| MDV | 784 | 12.25 | 18.1 | 85% |
| Midea | 682 | 12.5 | 18.0 | 84% |
| Mitsubishi Electric | 598 | 12.3 | 21.45 | 63% |
| TOSOT | 435 | 12.3 | 16.8 | 94% |
| GREE | 378 | 12.4 | 16.8 | 93% |
| BLUERIDGE | 351 | 12.2 | 18.0 | 92% |
| Cooper&Hunter | 302 | 12.5 | 18.0 | 82% |
| Lennox | 297 | 12.4 | 16.35 | 79% |
| MAINLINE | 233 | 12.3 | 17.4 | 99% |
| Carrier | 231 | 12.5 | 18.0 | 87% |
| Bryant | 228 | 12.5 | 18.0 | 86% |
| Innovair | 203 | 12.25 | 18.0 | 95% |
Look at the median column. Twelve of the largest brands in the category sit between 12.2 and 12.5. Across all 53 brands with 100 or more records, the full range is 13.0 (Kinghome) down to 11.85 (Haier) — under 10% from top to bottom.
Brand is a weak predictor of the typical unit. It is a much better predictor of two other things.
What brand does predict
Peak capability and cold-climate commitment
- The ceiling. Mitsubishi Electric reaches EER2 21.45 and COP 3.28 at 5°F — the highest of both in the entire dataset. Most brands top out around 16.8 to 18.0. If you want the flagship, brand narrows the search.
- Cold-climate share. This varies enormously and it is the most decision-relevant brand-level number on the page. MAINLINE certifies 99% of its range as cold climate, Innovair 95%, TOSOT 94%. Trane / Mitsubishi Electric is 54% and Mitsubishi Electric itself 63%.
- Refrigerant policy. Brands differ in how fast they have moved off R-410A (GWP 2,088) toward R-454B (470) and R-32 (675).
A high cold-climate share does not mean the units are better; it means the brand aims its range at winter heating. In a mild climate it is irrelevant. In a cold one it is the first filter. Why cooling efficiency barely predicts heating.
The finding
One model, ten badges
Filter the certified data to 9,000 to 18,000 BTU cold-climate units and sort by COP at 5°F. The top ten results are not ten machines.
They are one model — DLCPRBH18AAK, also filed as 38MPRBQ18AA3 — at 18,000 BTU, SEER2 26.3, EER2 15.3, HSPF2 11.00 and COP 2.89 at 5°F, appearing under:
Airquest, Arcoaire, Bryant, Carrier, Comfortmaker, Day & Night, Heil, Keeprite, Kenmore and LPS Legacy.
Every published figure is identical because it is the same machine. The pattern repeats across the dataset: Tempstar, SMART COMFORT and several others show the same maximum EER2 of 18.0 and the same maximum COP of 2.89, because they draw on the same platform.
It is the same structure we documented in room air conditioners, where 282 of 485 certified units share every published spec with another brand. Here the badge count per platform is simply higher.
The practical consequence is the same: where two candidate units have identical certified figures, they are the same machine, and price, warranty terms and local installer support are the only rational tie-breakers.
The label
Most Efficient does not narrow anything here
12,434 of the 14,328 certified units carry the ENERGY STAR Most Efficient designation — 86.8% of the field. Among the large brands the share runs from about 52% to 96%.
A label that most of a category holds is a floor, not a shortlist. That is the opposite of how selective the same label is elsewhere — 35.3% of room air conditioners, 14.7% of ceiling fans, 8.8% of bathroom fans, 0% of range hoods.
In mini-splits you have to read EER2 and COP at 5°F directly.
This one is not a DIY job
In this category the installer matters more than the brand. A mini-split’s published efficiency assumes a correct refrigerant charge, a properly sized and insulated line set, and an outdoor unit sited where it does not recirculate its own exhaust. Get any of those wrong and the figures on this page stop applying.
Charging, evacuating and brazing a line set is licensed work. Refrigerant handling is regulated.
Brand does affect one thing we cannot measure: parts availability and the local service network. A platform sold under ten badges may have very different support behind each one, and that is a legitimate reason to overrule this ranking.
Sizing should come from a proper load calculation, not a square-footage rule.
Say the unwelcome thing
Do not choose a mini-split by brand reputation.
The certified data does not support it. Across the 53 brands with 100 or more records, median EER2 varies from 13.0 to 11.85 — under 10% between the best and worst typical unit, in a field of 14,328 machines and 156 badges.
Meanwhile one 18,000 BTU model appears under ten different brand names with identical published figures, and the top ten cold-climate results in its size class are all the same machine.
Brand does predict two things worth using: the ceiling (Mitsubishi Electric holds the highest EER2 at 21.45 and COP at 3.28) and the cold-climate share (99% down to 54%). Use those as filters, then compare model numbers.
Read EER2 and COP at 5°F on the specific unit: EER2 19.1 and COP 2.48 or EER2 18.35 on R-454B.
In full
Every pick, and what is wrong with it
Published specifications only, each linked to the page it came from. Where a manufacturer publishes nothing, the row says so — we do not fill blanks in.
1. Best peak capability
Mitsubishi Electric Mini-Split Heat Pump, 9,000 BTU
Mitsubishi Electric · model MUZ-FX09NLHZ
Mitsubishi Electric holds the highest certified EER2 (21.45) and the highest COP at 5 degrees F (3.28) of any brand in the dataset.
The clearest case on this page of a brand whose ceiling and whose typical unit are very different things.
Highest EER2 in the whole certified field at 21.45, and the highest COP at 5°F at 3.28 — both the best of any of the 156 brands.
This model is SEER2 33.1, EER2 18.35, HSPF2 13.30, COP 2.20 at 5°F, on R-454B at GWP 470, cold climate certified and Most Efficient.
The catch is the range beneath it. Mitsubishi Electric’s median EER2 is 12.3 across 598 records, and only 63% are cold-climate certified against 99% for some smaller brands. Buy the model, not the name.
What’s wrong with it
Its median EER2 is 12.3 — below Midea's 12.5 — so the badge alone does not get you the flagship.
Running cost
Mitsubishi Electric does not publish a wattage for the MUZ-FX09NLHZ, so we cannot calculate what it costs to run. We are not going to estimate one. If Mitsubishi Electric publishes a nameplate figure, this block will show the arithmetic.
| Cooling capacity | 9,000 BTU/h |
|---|---|
| SEER2 | 33.1 |
| EER2 | 18.35 |
| HSPF2 | 13.30 |
| COP at 5 degrees F | 2.20 |
| Cold climate | Yes |
| Refrigerant | R-454B (GWP 470) |
| ENERGY STAR | Certified, and on the Most Efficient list |
| Noise | Not published in the certified dataset |
Every figure above is from ENERGY STAR Certified Mini-Split Heat Pumps dataset. Blanks are shown as “not published” rather than filled in.
2. Best median performer
Fujitsu Mini-Split Heat Pump, 9,000 BTU
Fujitsu · model AOUH09KTAP1
Fujitsu has one of the highest median EER2 figures among brands with 100+ records, and this unit is EER2 19.1 with COP 2.48 at 5 degrees F.
If you are going to shop by brand — which this page argues against — a high median is the property worth shopping for, because it is what you get when you pick at random.
Fujitsu’s median EER2 of 12.575 puts it in the top handful of brands with 100 or more certified records, above Midea, Carrier, Bryant and Mitsubishi Electric.
This unit is SEER2 33.5, EER2 19.1, HSPF2 14.00 and COP 2.48 at 5°F — the highest EER2 of any 9,000 to 24,000 BTU unit we hold data for.
R-32 at GWP 675. Lower than R-410A’s 2,088 and higher than R-454B’s 470, which is a real if secondary consideration.
What’s wrong with it
R-32 at GWP 675 rather than R-454B at 470, and it is a smaller range than the volume brands.
Running cost
Fujitsu does not publish a wattage for the AOUH09KTAP1, so we cannot calculate what it costs to run. We are not going to estimate one. If Fujitsu publishes a nameplate figure, this block will show the arithmetic.
| Cooling capacity | 9,000 BTU/h |
|---|---|
| SEER2 | 33.5 |
| EER2 | 19.1 |
| HSPF2 | 14.00 |
| COP at 5 degrees F | 2.48 |
| Cold climate | Yes |
| Refrigerant | R-32 (GWP 675) |
| ENERGY STAR | Certified, and on the Most Efficient list |
| Noise | Not published in the certified dataset |
Every figure above is from ENERGY STAR Certified Mini-Split Heat Pumps dataset. Blanks are shown as “not published” rather than filled in.
3. The ten-badge model
Carrier Mini-Split Heat Pump, 18,000 BTU
Carrier · model 38MPRBQ18AA3
18,000 BTU at EER2 15.3 and COP 2.89 at 5 degrees F — and the identical model number appears under at least ten brand names.
Included as the concrete proof of this page’s argument rather than as a brand recommendation.
Filtering the certified data to 9,000 to 18,000 BTU cold-climate units and sorting by COP at 5°F, the top ten results are the same machine — model DLCPRBH18AAK or 38MPRBQ18AA3 — under Airquest, Arcoaire, Bryant, Carrier, Comfortmaker, Day & Night, Heil, Keeprite, Kenmore and LPS Legacy.
SEER2 26.3, EER2 15.3, HSPF2 11.00, COP 2.89 at 5°F, cold climate and Most Efficient certified.
R-410A at GWP 2,088 is the honest drawback, and still about 38% of the certified mini-split field — what the refrigerant line means.
What’s wrong with it
R-410A at GWP 2,088, the highest-warming refrigerant still common in the category.
Running cost
Carrier does not publish a wattage for the 38MPRBQ18AA3, so we cannot calculate what it costs to run. We are not going to estimate one. If Carrier publishes a nameplate figure, this block will show the arithmetic.
| Cooling capacity | 18,000 BTU/h |
|---|---|
| SEER2 | 26.3 |
| EER2 | 15.3 |
| HSPF2 | 11.00 |
| COP at 5 degrees F | 2.89 |
| Cold climate | Yes |
| Refrigerant | R-410A (GWP 2,088) |
| ENERGY STAR | Certified, and on the Most Efficient list |
| Noise | Not published in the certified dataset |
Every figure above is from ENERGY STAR Certified Mini-Split Heat Pumps dataset. Blanks are shown as “not published” rather than filled in.
Skip this one
Hisense Mini-Split Heat Pump, 9,000 BTU
Hisense · model AO-09U3T25P
This unit is EER2 18.35 and cold-climate certified — while Hisense as a brand has one of the lowest median EER2 figures in the field.
Marked skip as a brand recommendation while remaining a perfectly good model — which is exactly the distinction this page exists to draw.
Hisense median EER2: 12.05. Among brands with 100 or more certified records, only AUX (12.0), GE Appliances (11.9) and Haier (11.85) sit lower.
This specific unit is EER2 18.35 with SEER2 32.0, HSPF2 11.80 and COP 2.25 at 5°F on R-454B — cold climate and Most Efficient certified. It is near the top of the field.
So “is Hisense a good mini-split brand?” has no useful answer. This model is good. Picking a Hisense at random is more likely to land near 12.05 than near 18.35.
What’s wrong with it
Hisense's median EER2 of 12.05 is near the bottom of brands with 100+ records.
Running cost
Hisense does not publish a wattage for the AO-09U3T25P, so we cannot calculate what it costs to run. We are not going to estimate one. If Hisense publishes a nameplate figure, this block will show the arithmetic.
| Cooling capacity | 9,000 BTU/h |
|---|---|
| SEER2 | 32.0 |
| EER2 | 18.35 |
| HSPF2 | 11.80 |
| COP at 5 degrees F | 2.25 |
| Cold climate | Yes |
| Refrigerant | R-454B (GWP 470) |
| ENERGY STAR | Certified, and on the Most Efficient list |
| Noise | Not published in the certified dataset |
Every figure above is from ENERGY STAR Certified Mini-Split Heat Pumps dataset. Blanks are shown as “not published” rather than filled in.
We are not linking a buy button for this one. If you want to look at it anyway, it is easy enough to find — but we would rather send you to something that fits your room.
Questions people actually ask
Frequently asked
What is the best mini-split brand?
The question has a weaker answer than you would expect. Across the 53 brands with 100 or more certified records, median EER2 runs from 13.0 down to 11.85 — under 10% between the best and worst typical unit. Mitsubishi Electric holds the highest peak figures; MAINLINE certifies the largest share as cold climate.
Do different mini-split brands sell the same units?
Frequently. One 18,000 BTU model appears under at least ten badges — Airquest, Arcoaire, Bryant, Carrier, Comfortmaker, Day & Night, Heil, Keeprite, Kenmore and LPS Legacy — with every published figure identical because it is the same machine.
Is Mitsubishi the best mini-split brand?
It holds the highest peak figures in the dataset — EER2 21.45 and COP 3.28 at 5 degrees F, both category records. But its median EER2 is 12.3, below Midea and Carrier at 12.5, and only 63% of its range is cold-climate certified. The flagship is exceptional; the typical unit is typical.
Does the ENERGY STAR Most Efficient label help choose a mini-split?
No. 12,434 of 14,328 certified units carry it — 86.8% of the field. A label that most of a category holds is a floor rather than a shortlist, so you have to compare EER2 and COP at 5 degrees F directly. How selective the label is.
What does brand actually predict in mini-splits?
Two things. The ceiling — what the brand's best unit can do, which varies from about 16.8 to 21.45 EER2. And cold-climate share, which runs from 54% to 99% and is the first filter that matters if you have real winters.
If two mini-splits have identical specs, which should I buy?
The one with better local installer support and warranty terms, then price. When every published figure matches it is the same machine with different stickers — and in this category installation quality affects real performance more than the badge does.
Sources
Where these numbers came from
Read next
Where to go from here
Back to Air Conditioners, or read how we pick.