Research count: 9 models weighed; 9 maker manuals among them; across this site's research, 497 sources kept in full; 148 Reddit threads read, kept as the reader's summary, not in full.
How the carpet figures were measured. Vacuum Wars buys its robots, embeds sand in medium-pile carpet, runs each robot at maximum power for a fixed time (usually three runs) and weighs the bin. Its pet-hair test flattens short pet hair into carpet and weighs what is left after two runs; its tangle test weighs long hair wrapped on the brushes after two runs on hard floor; its obstacle course counts objects avoided at maximum sensitivity. Its averages move as it adds robots, so each figure is judged against its own review's average. Consumer Reports states that robot scores compare only with other robots, and that the dividing line between robots and uprights is deep carpet cleaning.
Vacuum Wars results on every robot weighed for this page
Dreame D20 Plus. The 13,000 Pa claim sits against a bench suction of 0.53 kPa at Vacuum Wars, below that lab's average; Vacuum Chef (robot sent by Dreame, review states it was unsponsored) also measured below-average in-use pressure, and both note bench suction didn't predict its strong carpet results. Vacuum Chef measured 74 dB at maximum suction, slightly louder than average. The 91% pet-hair figure is the only weighed hair result found; no second lab has published one. Wirecutter's mop-combo testing calls it a light mopper that's prone to getting trapped and tangled, and estimates the bag at two to three months against Dreame's 150-day in-house claim. Just A Dad Approved's written three-month review rates its obstacle avoidance mid-tier and warns that the auto-empty path can clog with hair over time, but that review's text names the D20 Pro Plus, a sibling with a better front sensor, so we haven't leaned on it.
ECOVACS DEEBOT T90 Pro Omni. Vacuum Wars bought its own. Owner faults appear in Reddit threads where no owner confirmed the US model: a slipping brush-drive gear (three owners in one thread), fine water streaks, and detergent dosing that stopped. How often these happen is unknown. The thick-rug jamming comes from a three-month video sponsored by ECOVACS.
Dreame X60 Max Ultra Complete. RTINGS (bought its own in black; firmware 4.3.9_1495, July 2026) rated pet hair passable on a single default pass, found hard-floor pickup left notable fine and medium debris, and measured the dock wash water at 77 to 86 °F (25 to 30 °C). RTINGS' own threshold test stopped at 1.5 in (38 mm), and it notes tall thresholds must be marked on the map first; Vacuum Wars measured 2 in (51 mm). The Hook Up (unsponsored) lifted 65.5% of a mixed flour, salt, flax and rice load from high-pile carpet and found debris caked in the filter after auto-empty. Its edge-mopping test gave the X60 the top score of seven flagships. In its poop test it ran over the fake mess once before avoiding it on later passes. Just A Dad Approved saw the auto-empty clog once without warning during a packed pet-hair test, never at home. Carpets, debris and pass counts differ between these labs, which explains much of the spread; the heavy-shedder call rests on Vacuum Wars, AppleInsider's sponsored six-month home and RTINGS' pet-waste result.
Roborock Saros 20. Tom's Guide measured 70% of pet hair lifted from its own rug and 90% from hardwood in one run, and one cat owner reports hair left on very short carpet, so short, dense carpet may be a weak spot; that rests on two sources. The Hook Up lifted 62.8% of its mixed load from high-pile carpet. RTINGS' threshold test tops out at 1.5 in (38 mm), which it cleared. RTINGS and The Hook Up both saw it fail around pet waste, and George Langabeer's two robots (sent free by Roborock, the first on beta firmware) never avoided the fake poop.
Tapo RV30 Max Plus. TechGearLab measured 40% of real pet hair worked into carpet; Tom's Guide measured 67.8% on carpet with its own spread-hair method. Both agree it is the robot's weakest area. On the bag, TP-Link's US datasheet says up to two months; Wirecutter reports the company quoting three months elsewhere and estimates about two. Neither figure is a measured bag life.
Considered, not picked here. The eufy Omni C28 and Roborock Q7 M5+ are picks on other pages for other homes; for this reader neither recognizes obstacles well enough to beat the D20 Plus, and the Q7 M5+ is only average on carpet. For the iRobot Roomba Max 705 Vac, Consumer Reports' lab found it outstanding on pet hair and it passed CR's carpet test; Vacuum Wars hasn't tested it, and a three-month home test saw hair wrap at the brush ends and an auto-empty that didn't clear the bin first time after a carpet run. For the iRobot Roomba 105 Vac with its plain charging dock, the published lab figures are for the self-emptying bundle, not this robot, so they are left out; other pages on this site skip it.
Owner counts. Owner reports come from Reddit threads our research summarized. The counts are threads in a sample, not how common anything is, and in the topic sample no owner making a model-specific claim stated where they live. The heavy-shedder homes ran several different robots; they show the limit of robots in general, not of any one model on this page. One home with a heavily shedding dog found its replacement robot's dock emptied fine once the robot ran weekly instead of monthly, which kept each load within what the dock could pull.
Stairs. Four kinds of stair-climbing design were shown in 2025 and 2026: Roborock's legged concept, Dreame's tracked carrier, eufy's carrier shell and MOVA's prototypes. None had a US launch in our research.