There has been one pressing question about the Roborock Saros Z70 since it was announced at CES 2025: Could a robot vacuum with a robotic arm possibly be a practical purchase for a normal person?
The Saros Z70 takes the "robot" in robot vacuum to a whole new level. Past referring to the brains of the automated cleaning operation, "robot" also refers to a whole ass robotic arm that unfolds from the vacuum to move pesky obstacles out of the way. This is cool in theory for anyone whose home has random stuff lying around on the floor at any given time, but the practicality bubble is quickly burst when you realize that the Saros Z70 costs $2,599.
That's a full $1,000 more than Roborock's other two CES flagship releases, the Saros 10 and 10R, that I also happen to have at home right now. Multiple friends who have come over recently have literally laughed in my face when I told them that one of those three costs $2,599. And they're valid for that.
Naturally, I had to complete the Roborock Saros trifecta in my robot vacuum testing. The Saros Z70 has been living and cleaning in my apartment for almost a month now — the apartment whose monthly rent is half of what the Saros Z70 costs. In a time when most consumer electronics are getting more expensive because of tariffs, it's more important than ever to dissect whether these ultra-expensive robot vacuums are worth it.
First impressions of the Saros Z70
If you can't picture a robot vacuum with a "foldable five-axis mechanical arm" with a claw on the end, think of your average powerful robot vacuum but if it had limbs reminiscent of the robot dogs in that one episode of Evil, the adrenaline rush of a skating rink claw machine, and the harrowing sentience of a driverless Uber.
I will not elaborate further, but it reminds me of the giant scorpion from Clash of the Titans.
When the arm's not exposed, the Saros Z70 looks much more chill and chic than you might expect. My living room is currently lined with the Roborock Saros 10, 10R, and Z70, and they all look very similar in design. The Saros Z70 looks nearly indistinguishable from the Saros 10R that I just reviewed, from the mirrored dock exterior to the exposed water tanks atop. The fact that the Z70 is casually wielding a folded two-foot-tall arm inside is clever from a design standpoint.
But if I may pause the arm fanfare for a second, I think the actual cleaning performance is the biggest decision maker in a robot vacuum. So I'm getting into the Saros Z70's cleaning skills first.
Is the Saros Z70 good at cleaning?
As one of the most powerful robot vacuums on the market, yes, the Roborock Saros is a pretty thorough robotic vacuum cleaner.
Clocking 22,000 Pa of suction power, it ties for the strongest robot vacuum on the market — one of its direct competitors being the (less expensive) Roborock Saros 10. Measuring just over three inches tall, it's also one of the slimmest robot vacuums on the market. The other two Saros models are just as slim, but they don't have a mechanical arm folded inside them. So that's impressive. It misses being able to squeeze under the dining room credenza by a smidge, but easily fits under the TV stand and lip of the refrigerator.
But the Saros Z70's high suction power on paper hasn't left my hard floors completely spotless every time. Most dry cat food, most food crumbs, most dust bunnies, and most dirt or rocks tracked in by the door are swept up, but the Z70 always seems to leave behind some debris on the outskirts of its cleaning path.
Take the floor in front of one of my automatic litter boxes, for example. The mat in front of my Leo's Loo Too catches a lot of the litter tracking after my cats hop out, but a few specks almost always make it out. The bigger mess happens when I need to change the bag in the drawer and scoop fresh litter into the top. The Z70 did an OK job at sweeping up the most condensed part of the spill, but ended up pushing much of the litter pile out to the corners of the cleaning zone.
To be fair, all robot vacuums have trouble sucking up the litter dust that gathers right at the edge of the mat. The Dyson with a laser ads write themselves.
The Saros Z70 seems to have a better grasp on rugs than on hard floors — the opposite of the case with most basic robot vacuums. It has sufficiently combed the crumbs and random salad leaves off the living room rug from eating at the coffee table. Said coffee table is round and shaped like an acorn, and the S70 skirts around it nicely. The pile of the kitchen runner rug is a similar height, and the Saros Z70 rarely misses the crumbs that always gather there.
The 22,000 Pa of suction power was really flexed when vacuuming the faux fur Tumble rug in the hallway between bedrooms. It's the softest rug I've ever stepped on, and I was nervous to send a super powerful robot vacuum over something so nice and fluffy. The Saros Z70 was gentle enough while vacuuming to not pull out any fluff, but simultaneously vigorous enough to reach about 90 percent pet hair pickup on several occasions.
I even did the Shark PowerDetect cordless vacuum test on it — because that Shark empties itself every time, it starts each cleaning session with a clean slate. So, if I take it over a rug to test how well a robot vacuum did, I know just how much hair the robot vacuum missed. The Shark's canister was basically empty after cleaning where the Saros Z70 had just been.
The story wasn't as pleasant with medium-pile bath mats. Because they lift off the ground more easily, the Saros Z70 seemed too powerful to vacuum without loosening a thread or getting stuck on a sucked-up corner completely.
Mopping is the Saros Z70's cleaning specialty
The Saros Z70's automated mopping has fewer quirks than its vacuuming. I am so glad that Roborock opted to give the Z70 dual spinning mopping pads like the Saros 10R has. (The fact that it swapped those for one large, singular pad on the Saros 10 was a huge drawback for me.)
After much snooping on Reddit to get an idea of other people's opinions on the top robot vacuum features, I know I'm not the only one who consistently sees better mopping results from the two spinning pads versus one large one. I personally have found that they provide better angles and floor agitation to scrub drier stains. My favorite sub-ability of the rotating mop design is corner mopping: One of the Z70's spinning pads is attached via hinge and can swing out from under the vacuum to scrub along edges when the Z70 senses a wall. I can't sleep at night if I know that there's a dried sauce spill under the counter or dust and hair collecting in the corners of the bathroom, so I definitely need a robot vacuum that won't skip out on mopping those.
For sopping up bigger wet spills, spinning mopping pads like these aren't quite as good at the singular vibrating pads like the Saros 10 has. But the Saros Z70 still wipes up most small puddles, as long as you keep your expectations in check. If you spilled an entire carton of milk or something, the Z70's mopping pads would definitely just spread that around — but it successfully cleaned up small wine spills on a hardwood floor. A thin layer of wine was left over after the first pass, but the Saros Z70 returned on its own to make a few extra passes where it noticed a soiled spot.
The Saros SZ70's dock automatically dispenses detergent into the tank when it refills, so your floors will be a little more sanitized than they would being mopped with just water.
Does the Saros Z70 arm actually work?
The mechanical arm does pick up most items it claims to be able to. The problem is, it doesn't claim to be able to pick up a whole lot.
I sent the Saros Z70 into a room with a sock, slipper, or sandal lying on the floor at least 20 times. The arm didn't have a 100 percent pickup rate with any of those. Socks had the highest success rate, only being missed once or twice after being mistaken for "fabric" (not on the list of grabbable items). The sandals variety included a super lightweight pair of Birkenstock dupes (these were picked up every time), a legit pair of heavier Birkenstocks with the cork bottom (picked up few times, mistaken as a "carry-out bag" once, and as "footwear that cannot be picked up" once), and a pair of foamy Air Jordan slides (picked up sometimes, identified as "footwear that cannot be picked up" other times).
I've watched the Saros Z70 parade around for several minutes with its arm raised but nothing in its grasp, seemingly not doing much other than looking ominous and trying to scare the cats. (It doesn't work. They don't care.)
The feeling of watching the Saros Z70 fumble a simple flip-flop is reminiscent of minor road rage. Given how short the Saros S70's list of arm-compatible items is, you'd think it would at least be an expert on the things it can pick up. And, though the arm is obviously the paramount feature here, setting the arm up to work automatically instead of manually was a little convoluted. The setting, surprisingly, isn't the default.
Credit to it, the Saros Z70 did drop off almost every item picked up in the correct location. I designated socks and paper balls to go to the included Roborock trash bin by my bedroom door and for tidied up shoes to be put by the front door. This process is called "sorting," and you can make it automatic or select items to be moved on a case-by-case basis.
Now, I never said sorting was an overly expeditious process, either. After picking up an item, the Saros Z70 often drives around aimlessly for a minute or two to think about where it's going. I could pick up the shoe and unload the dishwasher in the two or three minutes it has taken to find the sorting drop-off. But if you're a busy parent with kids who constantly leave their shoes in the middle of the room, I guess the shoe being put away slowly is better than you having to do it yourself.
If there's an item in the path that the Saros Z70 isn't recognizing, there is a manual pickup mode where you can try to pick up the item claw machine style. The Saros Z70 has a camera that can stream whatever the vacuum's seeing to the Roborock app, and arrows will pop up that can be toggled to get the perfect pickup angle. The rush is similar to that of trying to win a stuffed animal at the arcade — until you remember that you're going after an item on the floor that you could have picked up in a second with your hand.
Though I'm disappointed at the Saros Z70's lack of dexterity, I'm not exactly surprised. My fellow Senior Shopping Reporter Haley Henschel first experienced it during a live demo at CES in January, where the Saros Z70 was almost acting... lost?
While some of the issues during CES were said to be related to the confines of that hotel, I've seen the Saros Z70 struggle to complete a mission in my apartment. It'll blurt out that it's "unable to reach the target" during a cleaning and return to the dock unprompted, or insist that "no item can be sorted" 30 seconds after I click on an item that it deems sortable in the app.
Is the arm really much better than small obstacle avoidance?
If the arm had more variety, maybe. But right now, the Saros Z70's mechanical arm does not drastically change the automatic cleaning experience. You know that episode of The Office where Kevin's trying to "save time" by cutting words out of his dialogue, but he just ends up confusing everyone? That's kind of the level of expectation vs. reality when it comes to Saros Z70's arm efficacy.
Arm aside, the Saros Z70's small obstacle avoidance is about 85 percent reliable. It uses StarSight 2.0 with 3D ToF (Time of Flight) technology for obstacle avoidance, just like the Saros 10R. StarSight was proposed as a slightly more precise alternative to the LiDAR laser system used by many robot vacuums with the circular tower on top (the Saros 10, for example). But I personally haven't noticed a huge change in accuracy either way. All three of my Saros units detect small objects accurately most of the time, but still struggle not to run over thin white cords, like iPhone or laptop chargers.
Is the Roborock Saros Z70 worth it?
No, you should not buy the Saros Z70 at this time. If Roborock follows through on updating the arm's software and expanding the list of grabbable items to items like shirts or pet toys, it'd be much more helpful.
But at this point in time, the arm has been so hit or miss for me that I'd probably just turn the feature off completely (if I weren't intentionally using it to test it). So then, you have a $2,599 robot vacuum with comparable small obstacle avoidance to a slew of significantly more affordable competing vacs. It's the same story with actual cleaning performance: The Saros Z70 has very powerful suction and advanced mopping techniques, but none of it does an exceptional enough job to cost that much. I actually think the $1,599.99 Roborock Saros 10R does a more thorough job.
The best way I can frame it is this: If I were a Pokémon trainer and I had the Saros 10, 10R, and Z70 in a Pokéball, I'd be saying "Saros Z70, I choose you" the least. In the month that I was testing all three of those simultaneously in my apartment, I wasn't opting for the Z70 unless it was for testing purposes. As someone who tests robot vacuums for a living, when I have multiple at my disposal at the same time, I think it speaks volumes when there's a certain robot vacuum that has performed so well in testing that I subconsciously reach for it over the others (or, in this case, don't reach for it).
