Buying guides
Home-fit finder
Buying guide
Verdict
Swept's pick for a mostly carpeted home is the Dreame X60 Max Ultra Complete: consistent in medium and high pile, a 2-inch climb, and mop pads it leaves in the dock.
BuyBest for most carpeted homes
Opens Amazon.com · Prices change often
At a glance
SuitsCarpet-heavy homes with deep rugs and a pet that drops heaps of fur, slips up now and then, or lives a step of more than an inch up
The catchRug tassels still tangle its brushroll
SuitsCarpeted homes with a sunken room or tall threshold and pets that never have accidents indoors
The catchCan't be trusted to avoid dog mess, least of all on carpet
SuitsCarpeted homes with shedding pets that don't have accidents and thresholds under an inch
The catchFailed a 1 in (25 mm) step at Vacuum Wars, and nobody has tested it on high pile
SuitsWood or tile homes with flat doorways, thick area rugs that must stay dry, a light shedder, and pets that never have accidents
The catchLeaves pet hair pressed into carpet
SuitsCarpet and rug homes where someone picks up socks and cords before it runs
The catchAvoids nothing small, so the floor has to be cleared first
Why you see a price band, not a price Prices change daily, so our pages show only the price band a robot sits in and leave the current price to Amazon. That way a page is never wrong by Tuesday.
Swept's pick for a carpeted home is the Dreame X60 Max Ultra Complete, and the 35,000 Pa on its box has nothing to do with it. It is the most consistent carpet robot across the labs we weighed, in medium pile and high pile alike. That is rarer than it sounds. The robot that collected the most on one lab's deep-pile test was rated poor on carpet by another. The X60 dug buried sand out of medium pile well above average, and RTINGS rated it good on low and high pile. A sunken den or raised family room won't stop it: labs saw it get up a 2 in (51 mm) sill. When it vacuums, its mop pads stay in the dock, so nothing wet touches your rugs. The catch is fringe: RTINGS found rug tassels still wind into its brushroll.
Pile height comes first. Most labs test in medium pile, the stuff in a typical bedroom. A robot that pulls grit out of it isn't always the one that lifts crumbs off a deep living-room rug. So we read the two apart, and checked each maker's own limit. eufy caps the Omni C28 at 0.7 in (18 mm) of pile and the Omni E25 at 0.8 in (21 mm). Roborock rates the Saros 20 to 1.18 in (30 mm), and eufy claims 2 in (50 mm) for the Omni S2. A rug over the limit gets avoided, or fought. A home whose rugs all stay under that low cap, on hard floors with a house-trained shedder, is a different reader; our pet-hair roundup picks the C28 there.
Then the step into the room: most robots still stall at about 0.8 in. Then dog mess. Every lab that compared found camera avoidance weaker on carpet than on wood. A robot that drives through it on a rug ruins a weekend. Then budget, where the best carpet cleaner for the money has no eyes at all. Mostly hard floors with a couple of thin rugs is a different job, and the home-fit finder will point you right.
BuyBest for most carpeted homes
Price band: under $1,700
Dreame sells the X60 on a carpet pressure plate, a self-lifting chassis and a climb of 1.8 in (45 mm). The climb claim is modest. Labs watched it clear 2 in (51 mm), the lip of a raised family room handled without you. Vacuum Wars works sand down into medium pile and weighs what comes back out; the X60 recovered 89%. On The Hook Up's deep-pile test of flour, salt, flax and rice, it collected 65.5%, close behind the leaders. No robot topped every test. The X60 was good in nearly all of them.
It lifts pet hair without wrapping its brush. Of the picks, it has the best record around dog mess, though labs found it surer on wood than on a rug. One month-long reviewer saw it treat dark rugs as rugs, not cliffs, which matters more than it should.
The catch is tassels. In a mostly carpeted house with plain-edged rugs, we'd buy it when your pet drops heaps of fur, slips up now and then, or has to climb more than an inch into a room. Where every threshold sits under an inch and the pet never has accidents, the L60 Ultra PE below costs a band less and is our home page's choice for that house; we agree. The X60 is also our pick for pet hair, and the full case is in our X60 Max Ultra Complete review.
BuyBest for long-coated dogs and tall steps
Price band: under $1,600
Roborock's spec sheet rates the Saros 20 for one 1.77 in (45 mm) sill. Then the robot outclimbs the paperwork. It cleared 2.2 in (57 mm) at one lab and 2 in (50 mm) both ways at another. One three-month tester saw it take a porch step of about 2 in every day. On carpet it matches the X60's 89% in medium pile, and RTINGS rated it good on high pile. It trailed on The Hook Up's deep-pile run, with 62.8%. Roborock rates its deep-pile mode for rugs up to 1.18 in (30 mm), and that mode only runs once you enable it in the Roborock app. Long hair doesn't wrap its brush, and the pads wait in the dock for carpet runs.
The catch is dog mess. RTINGS found you can't rely on it to avoid pet waste, and its avoidance is much worse on carpet. Two more testers watched it drive over their fake dog mess, which is why our roundup of robots that avoid dog poop skips it. What separates its reader from the X60's is the pets. We'd buy the Saros 20 for a carpeted home with a sunken room or tall threshold whose pets never have accidents indoors. Where a pet still slips up indoors, the X60 also clears a 2 in sill and is the better buy; see our X60 review. Our Saros 20 review has the detail, and Saros 20 against the X60 has the head-to-head.
BuyBest for wall-to-wall carpet on one level
Price band: under $1,000
Dreame's headline says the L60 Ultra PE climbs 1.57 in (40 mm). The fine print says that is two steps added together; a single step is 0.9 in (22 mm). Vacuum Wars watched it fail a 1 in (25 mm) board. In the factories I worked in, a figure quoted for two parts stacked together went back to the supplier.
Keep it on one level, though, and nothing here cleans carpet deeper. Vacuum Wars weighed 94% of its embedded medium-pile sand coming back out, and every strand of the flattened pet hair. None of it wrapped around the brush. Those numbers come from one lab; two long home tests back them up without measuring. Nobody has tested it on high pile, and one home tester saw it learn to steer around fluffy bath mats. Treat a deep rug as unknown.
The catch is the step. If your carpeted home has no threshold an inch or taller and a shedding pet that doesn't have accidents, we'd buy the L60, a price band below the X60. Our home page and our pet-hair roundup pick it for the same reader, and our L60 Ultra PE review makes the full case. A pet that has accidents, or a step an inch or taller, tips the call back to the X60.
BuyBest for rugs on hard floors
Price band: under $700
Roborock says the Curv 2 Flow's roller mop lifts on carpet and a shield drops over it. That claim holds up handsomely. In The Hook Up's run of seven flagships, it was the only robot to cross carpet mid-mop without wetting it. If your rugs sit on wood and must stay dry, that is the reason to buy. It is good on deep pile too: 67.3% on that test, ahead of the X60 and the Saros 20.
The catch is pet hair. RTINGS rated it poor at lifting hair pressed into carpet, because its brush doesn't dig into the fibers. It is also too tall for low couches and stops at a 0.75 in (19 mm) step. If your thick rugs sit on wood or tile, your doorways are flat, and your pets shed lightly and never have accidents, we'd buy it.
Change any of those and the answer moves. A pet that sheds hard onto deep rugs over wood or tile is the home our robot vacuum and mop roundup gives to the Dreame X60 Max Ultra Complete, which also reaches baseboards and toe kicks better; see our X60 review. A puppy or a pet that has accidents on mostly hard floors goes to the eufy Omni S2, the pick on our home page and in our roundup of robots that avoid dog poop. A sunken room or tall threshold with house-trained pets goes to the Roborock Saros 20, as our home page also says; see our Saros 20 review. Thin rugs on a mostly hard floor are a different job again, and the home-fit finder sorts that home out.
BuyBest on a budget
Price band: under $300
The Dreame D20 Plus claims 13,000 Pa, the smallest number of any pick. It still dug 93% of the buried sand out of medium pile, more than flagships claiming nearly three times the suction. It picked up 91% of flattened pet hair and wrapped none. Dreame says its dock bag lasts up to 150 days. Wirecutter's estimate is two or three months, which is still a long time not to think about it.
Everything else is honest budget. It has no carpet sensor, so you draw rugs on the map in the app. Its mop is a flat pad that drags, so testers leave it off. Dreame's manual limits it to a 0.8 in (20 mm) step, and nobody has tested it on high pile or fringe.
The catch is that it sees nothing small. Vacuum Wars laid out 24 obstacles and it dodged none, and Wirecutter found it prone to getting trapped. Anything small left on the floor is yours to pick up first. If someone in your house does that, we'd buy it. It is also our home page's vacuum-only pick and our vacuum-only pick for tidy homes, for the same reason.
SkipSkip for carpet; the Dreame X60 Max Ultra Complete suits carpet better
Price band: under $1,300
The S2 out-collected every pick here on deep pile and still loses. eufy says it handles carpets up to 2 in (50 mm). On The Hook Up's deep-pile run it collected 78.8%, against the X60's 65.5%, and Vacuum Wars measured 87% in medium pile. Then RTINGS rated it poor on both low and high pile, and bad on pet hair. We counted all three. Loose debris comes up well. Material pressed into the pile, which is what a lived-on rug holds, does not.
The mop settles it. The roller lifts but has no cover, and it wet carpet in The Hook Up's mixed-floor run. RTINGS also saw it stuck on tasseled rugs. It climbed 1.2 in (30 mm), short of the 1.4 in (35 mm) eufy claims. For a carpeted home we'd skip it and buy the Dreame X60 Max Ultra Complete. What separates this page's reader is the floor. On mostly hard floors with a puppy or a pet that has accidents, the S2 is the pick on our home page and in our roundup of robots that avoid dog poop, and our eufy Omni S2 review makes the case.
Two changes matter on carpet. Climbing chassis have arrived at the top end: three robots here cleared a 2 in step at the labs, where the old limit sat near 0.8 in. Roller mops are spreading too, but a roller doesn't promise dry rugs. The Curv 2 Flow kept carpet dry; the S2 wet it.
The newcomer to watch is the MOVA V70 Ultra Complete, released in August. Vacuum Wars measured 92% of buried sand in medium pile and a 2 in (51 mm) climb on a US robot. No other lab has measured the US model yet. One heavy-shedding owner found hair stalling at the dock's intake and the bagless bin filling fast, and Vacuum Wars found its mop water output erratic. Our pet-hair roundup and our other roundups call it a wait: one lab so far, a bagless dock unproven with heavy shedders, and mop water that needs a fix. We agree. For carpet we wouldn't hold out for it: the X60 has been measured every way that matters.
The research count for this page: 17 models weighed; 15 maker manuals among them; across this site's research, 497 sources kept in full; 148 Reddit threads read. None of the testing is ours. Every result belongs to the exact US model named. Where a lab tested a sibling, we say so; where nobody tested high pile, the page says not measured. Medium pile comes from Vacuum Wars' buried-sand test. High pile comes from The Hook Up's deep-pile run and RTINGS' ratings. Pile limits come from each maker's own guide or page. A pick needed results from at least two independent sources on the exact model.
No. The D20 Plus claims 13,000 Pa and pulled more buried sand from medium pile than the Saros 20, which claims 36,000 Pa. MOVA prints 42,000 Pa on the V70 Ultra Complete. At the brush, Vacuum Wars measured 0.6 kPa, mid-field. A pascal figure tells you about the motor the way a speedometer's top number tells you about your commute. The Hook Up puts carpet pickup down mostly to brush design, and the results here agree.
Read the box carefully: headline climbs are often two steps added together. Measured single steps run from the Saros 20's 2.2 in (57 mm) and the X60's 2 in (51 mm) down to 0.75 in (19 mm) for the Curv 2 Flow. The L60 failed at 1 in (25 mm). Testers who got the tallest climbs had marked the thresholds on the map first. No lab tests climbing a thick rug's edge on its own, so step height is the best guide we have.
The safest answer is a robot that leaves its pads in the dock, as the X60 and Saros 20 do for vacuum-only runs. The Hook Up found no spinning-pad robot lifted high enough to stop dragging on its carpet. The Curv 2 Flow's covered roller is the other answer that worked. The L60 lifts its pads rather than leaving them home. Dreame says it removes them for long-pile rugs; the lab that bought one recorded that it doesn't. The D20 Plus has a drag pad you take off yourself. The S2 lifts its roller high and still wet carpet. Want no water parts in the house at all? That's a different reader, and our vacuum-only roundup is written for them.
Almost every robot RTINGS ran over its tasseled rug had trouble. The X60 and the Curv 2 Flow tangled, and the Saros 20 shoved the rug around. The S2, the Saros 10R and the Matic got caught. The Q7 M5, the dockless twin of the Q7 M5+, got stuck almost immediately. The one clean pass came from the Shark AV2501S, which is weak almost everywhere else. eufy's C28 guide says to fold tasseled edges under. TP-Link's guide says not to run the RV30 Max Plus on fringed carpet at all.
It can. Most robots look down with infrared cliff sensors. Dreame's own explainer says black dyes absorb infrared light, so a dark rug reads as a drop. RTINGS found dark carpet can trip the cliff sensors on the Q7 M5, the dockless twin of the Q7 M5+. Roborock's Q7-series manual warns of reduced performance on dark carpets. One month-long reviewer saw the X60 read dark rugs as rugs, but nobody tests this robot by robot. RTINGS suggests covering the sensors. That also switches off the protection that keeps a robot from driving down the stairs. Dirty sensors cause the same symptom, and Roborock's Q7-series manual says to wipe them with a soft dry cloth if the robot starts missing spots.
| Robot | Buried sand, medium pile | High pile | Rug tassels | Buy |
|---|---|---|---|---|
| Dreame X60 Max Ultra Complete: most carpeted homes | 89% | Good at RTINGS; 65.5% at The Hook Up | Tangle the brushroll | View on Amazon (opens in a new tab) |
| Roborock Saros 20: tall steps, long-coated dogs | 89% | Good at RTINGS; 62.8% at The Hook Up | Pushes tasseled rugs around | View on Amazon (opens in a new tab) |
| Dreame L60 Ultra PE: wall-to-wall on one level | 94% | Not measured | Not tested | View on Amazon (opens in a new tab) |
| Roborock Qrevo Curv 2 Flow: rugs on hard floors | 87% | Okay at RTINGS; 67.3% at The Hook Up | Tangle and need help | View on Amazon (opens in a new tab) |
| Dreame D20 Plus: budget | 93% | Not measured | Not tested | View on Amazon (opens in a new tab) |
BuyBest for most carpeted homes
Most robots on this page are excellent on one kind of carpet. The Dreame X60 Max Ultra Complete stayed good from bedroom pile to a thick living-room rug across the labs that tried it. It climbs into the room a step up and keeps its wet pads off the rug. For a mostly carpeted home, it's the one we'd buy.
| Robot | Medium pile (Vacuum Wars, buried sand) | High pile | Maker's pile limit | Tassels | Dark rugs | Tallest single step any lab recorded |
|---|---|---|---|---|---|---|
| Dreame X60 Max Ultra Complete | 89% | RTINGS good; The Hook Up 65.5% | Not stated | Tangle brushroll (RTINGS) | Read as rugs (Jamie Andrews, 30 days) | 2 in (51 mm) Vacuum Wars; 1.5 in (38 mm) RTINGS |
| Roborock Saros 20 | 89% | RTINGS good; The Hook Up 62.8% | 1.18 in (30 mm), mode enabled in the app | Displaces tasseled rugs (RTINGS) | Not tested | 2.2 in (57 mm) Vacuum Wars; 2 in (50 mm) The Hook Up |
| Dreame L60 Ultra PE | 94% | Not measured | Not stated | Not tested | Not tested | Passed 0.75 in (19 mm), failed 1 in (25 mm) |
| Roborock Qrevo Curv 2 Flow | 87% | RTINGS okay; The Hook Up 67.3% | Not stated | Tangles, needs help (RTINGS) | Not tested | 0.75 in (19 mm) RTINGS; 0.5 in (13 mm) The Hook Up |
| Dreame D20 Plus | 93% | Not measured | Not stated | Not tested | Not tested | Manual limit 0.8 in (20 mm) |
| eufy Omni S2 | 87% | RTINGS poor; The Hook Up 78.8% | 2 in (50 mm) | Stuck (RTINGS) | eufy FAQ: dark carpet may not be detected | 1.2 in (30 mm) The Hook Up; 1.3 in (32 mm) once in three at Vacuum Wars |
| MOVA V70 Ultra Complete | 92% | Not measured on the US unit | Not stated | Not tested | Not tested | 2 in (51 mm) |
| Roborock Saros 10R | 80% | RTINGS decent | Not stated | Stuck (RTINGS) | Not tested | 1.5 in (38 mm) RTINGS; 1 in (26 mm) Vacuum Wars |
| iRobot Roomba Max 705 Vac | Not measured | Not measured on the US unit | Not stated | Not tested | Not tested | 0.7 in (19 mm) RTINGS |
| Roborock Q7 M5+ | 75% | RTINGS good (dockless Q7 M5) | Not stated | Stuck almost immediately (RTINGS, Q7 M5) | Can trip cliff sensors (RTINGS, Q7 M5) | 0.75 in (19 mm) (RTINGS, Q7 M5) |
| ECOVACS DEEBOT T90 Pro Omni | 81% | Not measured | Not stated | Not tested | Not tested | 1.3 in (32 mm) |
| MOVA P10 Pro Ultra | 81% | RTINGS acceptable | Not stated | Not tested | Not tested | 1 in (25 mm) RTINGS |
| eufy Omni C28 | 89% | Not measured | 0.7 in (18 mm) | eufy guide: fold tasseled edges under | Not tested | Claimed 0.8 in (20 mm) only |
| eufy Omni E25 | 88% | Not measured | 0.8 in (21 mm) | Chewed thin rug edges (TechGearLab); fringed rug tangled (Homes & Gardens) | eufy FAQ: dark carpet may not be detected | Claimed 0.8 in (21 mm) only |
| Tapo RV30 Max Plus | Not measured | Maker: not for high pile | Maker: not for high-pile carpet | Maker: not for fringed carpet | Not tested | Claimed 0.87 in (22 mm) only |
| Shark AI Ultra AV2501S | Not measured | RTINGS decent on rice and cereal | Not stated | Climbs them without often getting stuck (RTINGS) | Not tested | 0.7 in (19 mm) RTINGS |
| Matic | Well above that lab's average (no figure published) | RTINGS acceptable in Heavy Duty mode | Not stated | Caught (RTINGS) | No traditional cliff sensors; untested on dark rugs | 1 in (25 mm) RTINGS |
The Saros 20's brushes never needed cutting free in one home that ran it daily for 90 days with a shedding dog. That same home tester found the first map drew his rugs as circles, which broke the automatic pad drop until he traced the rugs by hand, a twenty-minute fix. The L60 Ultra PE also avoided 21 of 24 obstacles at Vacuum Wars, and the Qrevo Curv 2 Flow recovered 87% of buried sand in medium pile there. One more suction example: the Roborock Q7 M5+ claims 10,000 Pa and measured 2.24 kPa at the brush, several times the Saros 20, yet cleaned buried sand only as well as Vacuum Wars' average.
| Robot | Why it lost on this page |
|---|---|
| eufy Omni S2 | Skipped for carpet for the reasons above; on mostly hard floors with a puppy or accident-prone pet it is our home page's pick, and our review covers that buyer. |
| MOVA V70 Ultra Complete | Strong carpet numbers from one lab on the US unit, but an exact-model owner with heavy shedders found hair stalling at the dock intake, and Vacuum Wars found its mop water output erratic. Our pet-hair roundup and our other roundups call it a wait until a second lab tests it and the bagless dock proves itself with heavy shedders. |
| iRobot Roomba Max 705 Vac | Consumer Reports found it passed its medium-pile carpet test with flying colors, on a carpet list more than 12 months old, and no lab has published a carpet figure for it. One three-month home tester saw the dock fail to empty the bin after a carpet run on the first try. |
| Roborock Saros 10R | Middling on carpet, stuck on tassels, and labs split hard on pet hair. Our Saros 10R review, our pet-hair roundup, our roundup of robots that avoid dog poop and our under-$1,000 roundup all call it a wait, for mostly hard-floor homes with low thresholds and a light shedder, where dog mess on the floor is the worry. |
| Roborock Q7 M5+ | An average buried-sand result; its dockless twin got stuck on tassels almost immediately and can mistake dark carpet for a drop. Our home page picks it for small, tidy apartments, a different reader. |
| Tapo RV30 Max Plus | TP-Link's own guide says not to use it on high-pile or fringed carpet. We took the maker at its word; it belongs in small hard-floor homes, where our home page picks it. |
| eufy Omni C28 and Omni E25 | Strong in medium pile, but their makers cap rug thickness below what this reader owns. Hard floors and low rugs with a house-trained shedder are a different reader, and our pet-hair roundup picks the C28 for them. |
| ECOVACS DEEBOT T90 Pro Omni | A hard-floor robot with an 81% buried-sand result, and one sponsored three-month owner says it sometimes jams on a thick rug. |
| MOVA P10 Pro Ultra | RTINGS rated low pile poor and found its side brush doesn't run on carpet. |
| Shark AI Ultra AV2501S | The only clean pass on RTINGS' tasseled rug, but weak on fine grit and bad on pet hair there. |
| Matic | RTINGS found it poor on low pile and that at first it wouldn't drive onto carpet, while Vacuum Wars measured strong pet hair. Too split to recommend for carpet. |
Vacuum Wars buys its robots; its carpet figure is sand worked into medium-pile carpet and cleaned at maximum power, with the bin weighed, and its pet-hair figure comes from hair flattened into carpet and run twice. The Hook Up, unsponsored, in April 2026, spread a weighed mix of flour, salt, flax and rice on high-pile carpet and ran two vacuum-only passes at top suction; its auto-captions garble the Saros 20's name, so that robot is identified by its stated 36,000 Pa. RTINGS buys retail units, scores a sand, rice and cereal mix on low and high pile by eye, presses synthetic pet hair into low pile with a weighted roller, and runs a loose tasseled rug in its adaptability room. Its X60 review ran on firmware 4.3.9_1495 (July 2026), and its Q7 results come from the dockless Q7 M5, which RTINGS says is the same robot as the M5+. Several climbing robots top out at exactly 1.5 in (38 mm) at RTINGS, which looks like the top of its threshold range rather than the robots' limit. Consumer Reports tests robots on medium pile; its carpet list featuring the Roomba Max 705 dates from August 2025.
On the eufy Omni S2, The Hook Up's 78.8% on deep pile and Vacuum Wars' 87% in medium pile sit against RTINGS' "poor" on both piles. Method explains it: loose debris is lifted and pressed-in material is not. Vacuum Chef's deep coffee-grounds carpet test also scored the S2 below its predecessor. That unit came from eufy in late November 2025, ahead of the hardware upgrade eufy made before the S2's US launch, so the result may describe older hardware than the current S2. Why the brush behaves that way is still open.
The Dreame X60 has one weak lab result, CNET's sand on midpile, run by CNET's own method. RTINGS' tester said in comments he wouldn't recommend it "right now" because of software bugs, and one owner describes phantom walls appearing after a few days; firmware has also moved its carpet results upward at one lab. The Hook Up found 7.1 g caked in the bin after auto-empty, while another reviewer found only crumbs, and RTINGS' tester found the highest auto-empty suction setting helps a lot.
On the Roborock Saros 20, Tom's Guide measured 70% of pet hair on its own rug against 90% on hardwood, with an unclear dose, and a one-week cat owner on very short, dense carpet saw hair left behind. That may be a weak spot lab carpets don't reproduce; it is unconfirmed.
Dreame says the L60 Ultra PE removes its mops on long-pile rugs, while Vacuum Wars records that it doesn't drop pads at the dock. That stays open.
Where RTINGS' threshold figure sits below Vacuum Wars' and The Hook Up's on the same robot, RTINGS' test ceiling explains it, so the table gives the tallest single step any lab recorded.