CNET’s 2026 robot vacuum roundup put a new name in the overall slot on October 8: the iRobot Roomba Combo 10 Max, replacing the Mova V50 Ultra Complete. The number that got it there is 67.04 percent average sand pickup across floor types, with 73.62 percent on low-pile carpet. That is a lab result. It is not a reason to skip the bin, the pad, or the chair legs you still have to move.
If you already read our robot vacuum and mop guide, treat this as a score update, not a second taxonomy. The machines changed slots. The weekly work did not.
67 percent is sand in a lab, not your hallway
CNET’s overall pick is the Combo 10 Max because it led their sand average and because it cleared their obstacle course 6 out of 6. They say it was the first vacuum in their testing batches to do that, though two later units have since matched it. Coverage on that machine is 86.43 percent of the test room.
Sand on low pile is a controlled mess. Your hallway is cereal, hair, and the charging cable someone left by the sofa. A 67 percent sand average means the Combo 10 Max beat the rest of their board on that soil. It does not mean 67 percent of your Saturday is now optional.
The coverage figure is the one people misread. 86.43 percent of a test room is still a missed strip along the cabinet kick and whatever the robot decided was a cliff. If you have a runner that bunches, the percent on CNET’s heat map is not the percent on your floor.
Write the score on a note if you want. Then write the emptying interval next to it. Those are different columns.
The Saros 20 won coverage. That is a different trophy
CNET gave the Roborock Saros 20 their Lab Test award for cleaning coverage after it mapped 88.65 percent of the test room, replacing the Dreame X50 Ultra on that badge. Hardwood sand extraction on the Saros 20 is 82.48 percent. Midpile carpet pickup is 28.99 percent, close to the Combo 10 Max at 30.43 percent.
Yeedi’s C12 Plus posted 89.66 percent coverage in the same writeup, which is higher than the Saros 20, and CNET still handed the lab coverage award to the Roborock. Read the captions. Coverage and “best overall” are not the same row.
The Ecovacs Deebot T90 Pro Omni shows up as an honorable mention: 87.59 percent sand on hardwood, obstacle avoidance averaging four out of six and later five out of six, coverage 82.87 percent. CNET’s own caveat is that it is one of the better machines they tested if you are running it on hard flooring. If your house is mostly midpile, that sentence is the whole review.
A coverage award is useful when you are choosing between two docks that both claim they “go everywhere.” It is not useful as a substitute for walking the edges once a week with a stick vacuum, which is still how hairlines at baseboards get done. Our filter and brush maintenance note is the unglamorous half of that.
Pet hair is a second leaderboard
PCMag’s pet-hair roundup, dated October 6, is not CNET’s sand board. Andrew Gebhart’s notes on the Roborock Qrevo Curv are about two cats, a self-lifting chassis that climbs small thresholds, and a brush roll that did not tangle. The machine extends a side brush and mop arm into corners. PCMag listed it around $699.99 at Home Depot against a $1,599.99 strike-through.
That is a climbing chassis and a cat test. It is not CNET’s 6/6 obstacle course, and it is not a promise that the same unit will lead sand on low pile.
The Ecovacs Deebot X11 OmniCyclone sits on PCMag’s pet-hair page as a 4.0 with a three-hour consecutive cleaning run, bagless dock, strong large-particle pickup, and mopping they liked. Fine-particle pickup was weaker. Emptying dust from the base was tedious. One unit malfunctioned in testing.
Bagless sounds like less shopping. Tedious emptying is still a chore. If you buy the X11 because you hate bags, put “empty the cyclone” on the same weekday you already wipe the bathroom, not on a someday list. Consumer Reports scoring self-clean on an oven already made this point in a different room: the cycle is not the wipe.
6 out of 6 objects is not a toy policy
CNET’s Combo 10 Max avoided six of six objects. That sentence will sell docks this month. It should not retire the rule about charging cables, socks, and the dog bowl.
Object recognition is a lab course. The two later machines that also hit 6/6 prove the course is now a shared trick, not a personality. If a kid drops a crayon, you still pick it up before the 9 a.m. run. If you leave a fringe rug in the path, coverage percentages from a heat map will not argue with the robot when it eats the fringe.
Set the no-go lines for the things you already know are problems: the scale, the lamp cord, the bowl. Then let the 6/6 number do what it actually does, which is reduce how often you fish a USB cable out of a brush. That is a real gain. It is not a closed door.
What you still do on a weekday
Buy whichever dock matches the floor you actually have. Then schedule the parts the app will not brag about.
Empty the bin or the dock bag before it packs fine dust into a brick. CNET’s sand percentages assume the intake is not already full. PCMag’s X11 note that emptying the base is tedious is a warning, not a footnote.
Wash the mop pad if you enabled mopping. A pad that smells like the last kitchen spill will smear it. The Qrevo Curv extending a mop arm into a corner only helps if the pad is clean.
Pop the brush and pull hair. Pet-hair wins on PCMag’s page still wrap hair around a roller. That is the same five-minute job as a stick vacuum, just less often.
Check the dock’s dirty-water tank if the machine mops. A “self-clean” dock that sits on last week’s gray water is a humidifier for the next run.
Walk the edges the robot treats as cliffs: cabinet kicks, the stove toe-kick, the bathroom vanity. Coverage in the 80s is the middle of the room.
None of that is a product review. It is why a 67 percent sand score and a 88.65 percent coverage award can both be true and your kitchen still looks like Tuesday.
What to buy from this week’s lists
If your floors are mixed and you want the name CNET just moved into overall, the Combo 10 Max is the one they scored highest on sand and on the six-object course. Budget the iRobot consumables and the bin, not just the dock photo.
If your complaint is “it misses a strip,” look at the Saros 20’s 88.65 percent coverage award and at the Yeedi 89.66 percent figure in the same article. Coverage is the trophy. Midpile sand on the Saros 20 is still 28.99 percent. Do not buy coverage to solve carpet sand.
If the house is cats and thresholds, PCMag’s Qrevo Curv notes are the relevant test: climbing chassis, two cats, wall-hugging brush. Confirm the live price. $699.99 versus $1,599.99 on the same card is a sale, not a lifetime SKU.
If you want bagless and long runtime, the X11’s three-hour run and bagless dock are the features. Also budget the malfunctioned unit in testing and the tedious empty. Buy it only if you will empty it.
Hard floors only? CNET already told you the T90 Pro Omni is in that bucket.
Keep the old robot as a spare only if it still docks and you have a floor it already knows, like a basement. Otherwise sell it. Two maps of the same hallway is how hair stays in the corners while you debug Wi-Fi.
A week that uses the scores without worshipping them
Monday: empty the dock. If you bought the X11, this is the tedious bagless empty PCMag already complained about. Do it over the trash, not over the sink.
Tuesday: mop pad in the wash if the machine mopped kitchen. A Combo 10 Max that also mops will happily redraw last night’s sauce if the pad is still wet with it.
Wednesday: pick up the object course you already know you will fail: cables, socks, the scale. CNET’s 6/6 is the robot’s course, not yours.
Thursday: brush roll. Pet hair from PCMag’s two-cat test still wraps. Five minutes.
Friday: run the robot while you cook, then walk the toe-kicks the coverage heat map will never love.
Saturday: stick vacuum or broom on stairs and upholstery. No roundup this week tested those.
Sunday: check the app for a missed room instead of buying a new SKU because Saros 20 passed 88.65 percent somewhere else.
If you are replacing a Mova V50 Ultra Complete because CNET moved the overall crown, read the Combo 10 Max scores again before you click. 67.04 percent sand average and 73.62 percent on low pile are the reasons they moved it. 86.43 percent coverage is not better than the Saros 20’s 88.65 percent. You are buying pickup and obstacle avoidance, not the coverage trophy. If pickup was never your complaint, you are shopping because a headline moved.
PCMag’s $699.99 Qrevo Curv listing is the only price in these two pieces that looks like a normal dock. Confirm it at checkout. Strike-through $1,599.99 is a sale graphic. If Home Depot is $1,200 when you load the page, you do not have the deal they printed.
If that list looks like you do not own a robot, good. The robot is the middle of the room. CNET and PCMag are arguing about which middle is least bad. Your job is the rest.
Do not buy a second robot because the overall crown moved. One dock, a weekday empty, a pad wash, and a brush pull will beat a second map of the same hallway. The crown will move again. The bin will not empty itself.
