R290 heat pumps in 2026: which brands are paying the efficiency premium?

How big is the R290 versus R32 universe in 2026?

The surprise in the 2026 EPREL snapshot is how small the propane pool still is: R32 accounts for 13,935 declared models versus 537 for R290, or roughly 84.7% for R32 against 3.3% for R290 among declared refrigerant entries, a gap of 13,398 models (refrigerant_universe).

That matters because any “R290 premium” is being inferred from a much narrower slice of the market than the mainstream R32 catalog or even the broader heat-pump catalog. The declared refrigerant counts also show legacy HFCs have not vanished from registrations: R410A still appears on 1,896 models, plus another 49 entries as “R410a” and 10 as “R410”, taking the R410 family to 1,955 entries if those spellings are grouped, still far above R290’s 537 declared uses (refrigerant_universe). By contrast, other refrigerants are fringe in this snapshot: R134A appears twice, while codes like R23, R420A, R423A, R421A, R411A, R35, R422A, R413A, R425A, R419A, R33, R417A and R332 appear once each (refrigerant_universe).

So the declared usage table says the market is not simply “R32 versus R290”. It is still overwhelmingly R32-led, with a sizeable tail of older R410-family declarations and only token counts for everything else. Readers can cross-check the mix against the refrigerants reference and filtered R290 listings.

Does R290 actually buy higher efficiency in the aggregate?

The corpus does not provide a refrigerant-level average SCOP for all R290 models versus all R32 models, so the article cannot state the overall average SCOP gap between those two refrigerants with numbers from this dataset.

That limitation is important. The seed question asks for the average SCOP of all R290 models versus all R32 models and the absolute gap, but the available probes include declared refrigerant counts, type averages, brand averages, and an empty top_models for R290 response — not a direct refrigerant-by-SCOP aggregation (refrigerant_universe) (type_efficiency) (brand_share) (top_models).

What the corpus does show is that the broad European heat-pump market is dominated by categories where efficiency levels vary materially by type and by manufacturer. Across all listed products, air-water models number 30,452 with average SCOP 4.54, while ground-water models number 213 with average SCOP 4.77 and water-water models number 31 with average SCOP 6.15 (type_efficiency). That already suggests why a raw refrigerant comparison can mislead: if one refrigerant is concentrated in higher-performing product families or premium brand lines, some of the observed edge will come from composition rather than refrigerant chemistry alone.

For methodology-minded readers, that is the core caution in any refrigerant-efficiency narrative, and it is why our methodology page and market index matter as much as headline averages.

Type matters: the air-water comparison most readers care about

The most policy-relevant and buyer-relevant segment is air-to-water heat pumps: 30,452 models with an average SCOP of 4.54 and average output of 11.83 kW (type_efficiency). But the corpus still does not provide the air-water-only average SCOP split between R290 and R32, so it cannot confirm numerically whether the refrigerant gap persists within that type alone.

That absence matters because air-water is where many readers assume propane has already taken over on efficiency. The current dataset cannot verify that directly. It can only show that air-water dominates the space in which the refrigerant transition is happening, while higher-SCOP ground-water and water-water products remain tiny by model count: 213 and 31 models respectively (type_efficiency).

A clean same-type comparison would need an EPREL aggregation of air-water + refrigerant + SCOP. Without it, the best we can say is that type mix is a live confounder, and any premium attached to R290 should be treated as partly a model-selection effect until that split is available. Readers looking for current performance leaders can use the air-water SCOP leaderboard rather than assuming the refrigerant label answers the question on its own.

Which brands are overperforming or underperforming the refrigerant trend?

What the brand data does show is that the efficiency story is strongly brand-driven. Among the top 20 manufacturers by model count, the highest average SCOP belongs to LG Electronics Deutschland GmbH at 4.93 across 405 models, followed by STIEBEL ELTRON GmbH & CO. KG at 4.84 across 433 models, Bosch Thermotechnik GmbH at 4.69 across 3,602 models, GORENJE GOSPODINJSKI APARATI D.O.O. at 4.67 across 440 models, Ariston SpA at 4.66 across 2,618 models, and Gree Spain Corporation SL at 4.65 across 639 models (brand_share).

That ranking is notable because the biggest-volume players are not the efficiency leaders. Daikin Europe N.V. is by far the largest with 14,668 models and 24.05% share, but its average SCOP is 4.44 (brand_share). Mitsubishi Electric Europe B.V. holds 5,575 models and 9.14% share at 4.51 SCOP, while Hitachi’s European entity has 5,207 models and 8.54% share at 4.18 SCOP (brand_share). By contrast, Bosch Thermotechnik GmbH combines scale and above-market efficiency at 4.69 across 3,602 models (brand_share).

That supports the article’s angle more than a simple refrigerant argument does. Efficiency premiums appear to be concentrated in manufacturer portfolios, not just refrigerant choice. A buyer comparing Daikin Europe N.V., Mitsubishi Electric Europe B.V., Ariston SpA, or Vaillant GmbH should expect brand-level performance dispersion before even getting to refrigerant selection (brand_share).

What the corpus cannot answer is which of these large brands are R290-heavy versus R32-heavy, or each big dual-refrigerant brand’s exact model count and average SCOP by refrigerant. That probe is not present.

The best R290 models versus the best R32 models

The corpus cannot name the top individual R290 models or compare them with top R32 models in matched capacity bands, because the top_models probe for R290 returned no rows (top_models).

So the dataset does not support any numeric claim about the best individual propane unit, the best individual R32 unit, or the efficiency spread inside R290 models themselves. It also cannot support the question of whether the best R290 model is materially ahead of the median R32 model, because neither the top R290 values nor the R32 median is included in the available corpus (top_models).

The practical implication is straightforward: any article claiming exact “best R290” model names or point-by-point SCOP comparisons from this snapshot would be extrapolating beyond the evidence here. For product-level browsing, readers are better served by the top SCOP leaderboard, the air-water leaderboard, and the filtered R290 catalog.

What the 2026 EPREL data says about the refrigerant transition

The strongest 2026 takeaway is not that propane has already won on measured efficiency, but that the refrigerant transition is advancing from a very small installed choice set in EPREL: 537 declared R290 entries versus 13,935 for R32 (refrigerant_universe). That alone cautions against reading too much into anecdotal “R290 superiority”.

The second takeaway is that market structure still matters more than slogans. The biggest type by far is air-water at 30,452 models and 4.54 average SCOP, while the highest-performing types by average SCOP are tiny niches by count (type_efficiency). On top of that, brand averages range from 3.93 for Toshiba Carrier Europe S.A.S to 4.93 for LG Electronics Deutschland GmbH within the top-20 volume manufacturers, a full 1.00 SCOP point spread (brand_share). That is large enough to swamp many simplistic refrigerant narratives.

So the 2026 EPREL evidence available here supports a narrower claim: R290 is an important transition refrigerant with growing visibility, but the observed efficiency premium is likely to come as much from which brands and model lines adopt it as from refrigerant choice alone. To prove a clean refrigerant effect, the next missing piece is a refrigerant-by-type-by-brand SCOP cut from EPREL.

Sources

refrigerant_universe — IPCC AR6 GWP table; EU Reg. 2024/573 phase-out schedule; EPREL declared codes. Snapshot: 2026-07-20.
type_efficiency — EPREL Public API · type aggregation. Snapshot: 2026-07-20.
brand_share — EPREL Public API · brand-share aggregation. Snapshot: 2026-07-20.
top_models — EPREL Public API via Househeating Pulse catalog. Snapshot: 2026-07-20.