Best for:

The Mova V50 Ultra Complete is best for those seeking the strongest overall cleaning performance across flooring types. Its 65.14% sand pickup score across flooring types was the highest we saw in this batch of 24 robot vacuums. Homes with primarily midpile or plush carpets will also benefit from using it. The sand pickup test on midpile carpet was unprecedented, scoring 47.54%, which is 86% better than the runner-up. It's also suitable for people who value a lightweight base station for easy transport and a robot vacuum with a sleek design and good edge cleaning capabilities. All this earns it our Editors' Choice.

Why we recommend it:

We recommend the Mova V50 Ultra Complete because it offers exceptional cleaning performance where it matters most: extracting sand from carpets, with the highest overall average sand pickup score of 65.14%. It also excelled at low-pile carpet, coming just shy of the Eureka J15 Pro Ultra at 60.29% pickup.
It offers the third-best area coverage we've tested, with 80.78% area coverage in our test room, putting it just under the Dreame X50 Ultra and Roomba DustCompactor 205 Combo. The pop-out arm and roller brush both allow more comprehensive edge cleaning, which is a nice touch.

It successfully avoided common obstacles during our tests, including pet waste, socks and pet toys, and it has an accurate carpet suction boost. The retractable turret and leg, along with a three-year warranty and 30-day trial, make it a compelling and feature-rich choice. It's also about $100 cheaper than a comparable model, the Dreame X50 Ultra, when neither is on sale.

Who shouldn't get it:

This robot vacuum might not be ideal for those who aren't comfortable with initial setup difficulties, as the firmware update failed three times and took about 20 minutes. The device also initially failed to create a map (though it did after the first cleaning cycle). Also, while its intelligent obstacle avoidance had some successes, it still hit three out of six test objects. If you're expecting flawless navigation, you might be disappointed. Finally, its slow movement, while beneficial for cleaning efficiency, means a longer run time, which could be a drawback for those who prioritize fast cleaning cycles.

The $1,399 price is also on the higher end, so if budget is a concern, you may want to take a look at the Roomba 205, which runs you under $500 when not on sale. For superior obstacle avoidance and useful carpet cleaning features, we like the Eufy E28.

Best for:

The Dreame X50 Ultra is best for those who want comprehensive room coverage and excellent hardwood floor cleaning. It's also a robot vacuum designed to solve the problem of crossing thresholds. It can cross them, and obstacles up to 2.36 inches, by deploying a pair of small legs to lever itself across. It offers excellent vacuum performance on hardwood at 98.54%, nearly identical to that of the Roomba 205. And it has the best room coverage at 83.8%, cleaning the largest area of our test room compared with other robot vacuums. This earns it our lab award for cleaning coverage.

Why we recommend it:

In our lab testing, it successfully climbed a stack of plywood up to 2.36 inches, although it faced some issues coming down from anything more than 1.5 inches. Dreame says that it works best when there's a transition between the two vertical obstacles, such as a 1.65-inch threshold followed by a 0.7-inch threshold.

In cleaning capabilities, it's no slouch either. In lab testing, it removed 98.54% of sand from a hardwood floor. It also did well on carpet, with 51.88% pickup on low-pile and a midpile score of 25.51%. Mopping functionality is also solid. Our lab engineer took it home and tested the mopping and vacuuming on his 1,200 square-foot main floor. It mapped the main level in 11 minutes and cleaned everything in a little more than 2 hours. The base station also comes with two tanks, one for clean water and one for dirty water, along with another compartment for cleaning solution.

In real-world testing, it worked great with pet hair, removing most -- if not all -- of the pet hair from hardwood and carpet. The tangle-free roller brush never got tangled in a household with a Great Pyrenees. Navigation was also good, with the X50 Ultra able to identify objects such as shoes, socks, tennis balls, electrical cords and other items it could have gotten stuck on. It also recognized and worked around pets.

Who shouldn't get it:

At $1,600 at full price, it's one of the more expensive robot vacuums you can buy, so we only recommend it if you truly have numerous thresholds and challenging transition strips in your home. It also didn't last long, using up 20% of its charge during a single cleaning session for our relatively small 182.8-square-foot test room.

Best for:

The Roomba 205 DustCompactor Combo is the best for those on a tight budget who don't want to compromise too much on cleaning performance. It's our favorite midprice robot vacuum at $470 full price. It offers strong suction on hard flooring with an impressive 99.27% pickup, the highest we've tested and a total average of 58.64%. It also supports mopping and has a unique mechanical paddle that compresses debris in the dustbin, allowing you to store up to 60 days' worth of dirt in the robot vacuum at a time. You also get impressive area coverage, with our second-highest score at 82.44%, putting it just behind the Dreame X50 Ultra. All this makes it a great affordable pick for most people and earns our Editors' Choice.

Why we recommend it:

The 205 DustCompactor Combo gives you everything you need in a surprisingly small package. Our lab technician, Schylar Breitenstein, noted that the entire unit, including the docking station, is tiny, making it a good option for people in apartments or small homes. In terms of performance, it did well in our testing, scoring our highest hardwood score at 99.27%. It's not as strong on carpet, at 42.32% on low-pile and 16.23% on midpile, but it works out to an average of 52.61%, which we consider good.

As mentioned, area coverage is excellent, with our second-highest score. You can see on the heat map, there are very few missed spots. It actually does better than our best overall, the Mova V50.

It also mops, though strangely, the mopping settings aren't available in the app. Despite that, it still worked perfectly, taking about 30 minutes to clean our test area and removing all stickiness from the floor after the first cleaning cycle.

Who shouldn't get it:
Despite having a charging base station, the 205 DustCompactor doesn't self-empty or self-refill its water tank, so you'll still need to do that manually. If this is a deal breaker, many robot vacuums on our list have this feature, though they'll all run you quite a bit more. Object avoidance is also a disappointment. It avoided just one out of six objects, with two critical failures, where the test had to stop due to poop wrapping around the wheel and a sock getting sucked in. The only obstacle it successfully avoided was the 90-degree pet waste. For better object avoidance at a reasonable price, we like the Eufy E28, though it won't have as good of a cleaning performance on hardwood.

Best for:
The Roborock Saros Z70 is best for homeowners with modern, low-profile furniture who struggle to clean hard-to-reach areas. Because it is designed without a navigational turret on top, it was the only robot vacuum in our testing to successfully pass under our 3.5-inch clearance threshold. It is also a strong choice for homes with complex layouts and high transition strips, as its adjustable chassis allows it to climb over obstacles that trap other bots. The mechanical arm, while a neat feature, is ultimately not what earns it a spot on this list.

Why we recommend it:
While many robots get stuck under couches or beds, the Z70 navigates strategically around chair legs and glides under low gaps to clean where dust bunnies accumulate. In terms of raw cleaning power, it's a capable machine on hard floors, removing 89.05% of sand in our tests. It also posted a very high room coverage score of 83.27%, meaning it doesn't leave many patches of the floor untouched.

The Saros Z70's main selling point is the mechanical arm, which can pick up and sort small objects like crumpled papers, napkins, slippers and shoes. It's neat to watch the arm automatically emerge from the robot vacuum's body to do some pickup. It's fairly easy to designate a sorting zone in the app and the arm will usually put things where it's told but to be frank, the arm is hit-and-miss. It's capable of recognizing the objects it's trained for, but it usually wasn't able to pick up things like pet toys and, on one occasion, it confused my standing desk mat with a sock and got stuck trying to pick it up. (That's something Roborock is working on, though, and that potential makes it a robot vacuum worth watching.)
"The arm is a neat feature to watch, but I personally don't find it very useful," said Jared Hannah, senior technical project manager at our Louisville testing lab. "I feel like if the robot is smart enough to avoid obstacles such as shoes and socks, I don't need it to try and pick them up and move them. As long as the robot doesn't get stuck on anything during its cleaning cycle, that's good enough for me."

I was also impressed by its mapping. After a quick scan, the robot did a great job of generating a layout of my apartment, identifying flooring types, key pieces of furniture, pet areas and cords. Tagging things manually was also fairly straightforward, though if I have one gripe, it's that the app's settings and customization options are a bit overwhelming.
For me, the AdaptLift chassis was the real lifesaver. Like the Dreame X50 Ultra, it can deploy a pair of legs to lever itself over obstacles. It's one of the few robot vacuums that can cross the 2-inch thresholds in my apartment.

Who shouldn't get it:
If you have shedding pets like me, you should avoid this model. Our testing found its pet hair pickup performance to be terrible. Additionally, the object avoidance needs work for the price: It identified and avoided the pet toy and lamp cord, but failed to avoid the sock and two types of simulated pet waste. This also means the robot arm had trouble identifying objects as well. Finally, at $2,600, it is difficult to justify if you have significant carpeting, as it only removed 31.88% of debris from midpile carpet and totally missed sections of the rug during heat map testing.

Best for:

The Ecovacs Deebot X8 Pro Omni is a powerhouse best suited for pet owners who struggle with hair wrap but don't have pets prone to accidents inside the house. Its "zero tangle" claim held up in our lab, with the vacuum successfully clearing pet hair from hardwood and low-pile carpet without clogging the brush roll. It's also an excellent choice for homes with low-profile furniture, as its retractable lidar turret allows it to squeeze under 4-inch gaps that would trap standard robot vacuums.

Why we recommend it:

In terms of raw cleaning power, the X8 Pro Omni is a top-tier contender. It secured a high hardwood sand pickup score in its test batch at 97.08% and maintained a strong overall average of 60.28% across all floor types. We were particularly impressed by its intelligent design features. Unlike some competitors, its carpet suction boost worked reliably, and the retractable turret is a legitimate problem-solver for cleaning under beds and sofas. It also navigated safely around common household hazards such as a lamp cord, a sock and a pet toy during our avoidance tests.

Who shouldn't get it:

Who shouldn't get it: If your pets aren't house-trained, this robot poses a significant risk. During our obstacle avoidance testing, the X8 Pro Omni failed to avoid the 90-degree and 180-degree simulated pet waste. Worse, one of the fake messes got stuck in the main brush.

"One of the simulated pet poops got stuck in the dustbin main brush," said Breitenstein. "This would’ve been catastrophic if it were actual pet poop." Additionally, if your home is primarily covered in plush or midpile carpet, you may find the cleaning results inconsistent, as the vacuum's pickup rate on this surface fluctuated wildly (from 10% to 36%) during our five test runs.

Best for:

The Yeedi S20 Infinity is the best choice for eco-conscious folks, or for those tired of paying for disposable vacuum bags. Because it uses a clear, washable dust reservoir in its base station rather than a throw-away bag, it's a money-saver in the long run. It's also an excellent pick for homes with predominantly hard flooring and shedding pets, as it achieved near-perfect sand removal on hardwood (94.89%) and handled pet hair without tangling.

Why we recommend it:

We recommend the S20 Infinity primarily for its base station design, which our lab team highlighted as a standout innovation. "The new OmniCyclone base station is an amazing engineering development for the robot vacuum world," said CNET lab technician Schylar Breitenstein. "This completely erases the need to buy replacement bags and breaks down an unnecessary paywall."

Beyond the base station, it is a capable cleaner on hard surfaces, removing 94.89% of sand in our tests. It also proved to be low-maintenance regarding hair wrap. Our testing showed "no hair left in or around the dustbin after the pet hair test". Navigation was generally safe, with the unit avoiding the sock, pet toy and two types of simulated pet waste.

Who shouldn't get it:

If your home has wall-to-wall carpeting, specifically plush or midpile rugs, this robot is not for you. It struggled significantly on softer surfaces, retrieving only 23.19% of debris from midpile carpet. Breitenstein also noted that the unit "didn’t accurately use suction boost on carpet like Yeedi says it does," which likely contributed to the lower scores. Additionally, while it avoided most objects, it failed to avoid a lamp cord and one configuration of pet waste, so it requires a slightly tidier floor than the top-tier obstacle-avoidance models.

Best for:

The Eufy Omni E28 is the best robot vacuum for impressive obstacle avoidance and portable spot cleaning functionality. It's another of Eufy's unique three-in-one robot vacuums. Unlike the E20, which incorporates a cordless and handheld vacuum into the robot, the E28 is a portable deep cleaner built into the robot's base station. The detachable dual water tank (clean and dirty water) doubles as a deep cleaner, complete with a handle and hose attachment that lets you spot-clean various soft surfaces, including carpets and fabric.

"At home testing the carpet cleaner, I loved the self-cleaning feature on the extractor," said Breitenstein. "It cleans very well. It does need to be plugged in to use, but with an extension cord, it's very attainable to clean everywhere." Add excellent obstacle avoidance that managed to avoid five of six obstacles, and you get a very capable robot vacuum for $1,000 full price, though it's usually less on sale.

Why we recommend it:

In terms of performance for regular usage, the robot vacuum scores on the average side, with 44.49% sand pickup across flooring types. On carpet, it gets 40.58% from low-pile carpet and 11.88% from midpile, which isn't as high as we'd have liked for something specifically designed to clean carpets. Obstacle avoidance was also excellent, with five of six obstacles avoided.

In terms of other capabilities, it has all the key elements you'd expect, including self-washing and drying for the mop, brushes that prevent entanglement with pet hair, the ability to clean corners and self-emptying and automatic refilling for the base station.

At full price, the E28 will run you $1,400, but you can often get it on sale for less, which makes it a good option if you have a lot of carpeting at home but need to save space by not having a dedicated carpet cleaner. "For $999 or less, I think this price is completely justified considering it comes with a carpet cleaning function," said Breitenstein, also noting that the dustbin is quite large compared to competitors.

Who shouldn't get it:

If comprehensive room cleaning is important to you, the E28 doesn't do so well. It had the lowest average cleaning score at just 70.03% -- the worst of any robot vacuum we tested in this batch of 12. As the heat map shows, it missed a significant number of spots, left lines between cleaning areas and wasn't able to fit under two items of simulated furniture. For superior cleaning coverage, we like the Dreame X50 Ultra (83.8%), Roomba 205 (82.44%) and Mova V50 (80.78%).