ECOVACS DEEBOT X11 OmniCyclone robovac/roller mop lets the good time roll (review)

The ECOVACS DEEBOT X11 OmniCyclone robovac/mop was released at IFA 2025, and we have the first Australian unit for review. It has been through its paces, and our testers give it a ‘tick’.

Our seminal Five tips for choosing a robovac/mop (2025 cleaning guide) covers all the generations from the Gen 1 bump and grind cheapie to what we call Gen 5, capable of unattended whole-of-home cleaning. Well, it’s now about time to introduce the Gen 6 for this new crop of AI-driven, roller mop, robot vacuums.

Why? AI takes the guesswork out of settings (how to clean a floor type better), it’s better at navigation and obstacle avoidance, and as it gets to know your home, it cleans it better.

Our first robovac roller mop reviews in 2024 showed some potential, but they were not quite there yet. Excellent at mopping, but the AI was not mature.

• Eufy S1 Pro – the robot vacuum/mop that rewrites the book (cleaning review)
• ECOVACS DEEBOT X8 PRO OMNI – new extendable roller mop (cleaning review)

Now we have ECOVACS, eufy, Roborock, Narwal, Dreame, Mova, and more wanting us to review the next generation.

Roller versus platen or rotating mops

If you don’t know, robot vacuum/mops typically have a static/vibrating platen or dual rotating mop pads. To be clear, these quickly become soiled, requiring a wash every 15-20m² to avoid dragging dirty water over the floor. That is not to say that these are inferior – it depends on your floors and if you frequently drop a cup of tomato sauce on them😁😮.

Roller mops rotate around 200 RPM, and clean water wets the roller while squeegees remove wastewater and dirt. Simply put, you are not dragging a dirty mop over the floor. That is a moot point if you are not wont to spill tomato sauce (or any other such viscous liquid, which you should hand wipe up anyway) because all three types do a decent maintenance mopping job.

However, ECOVACS has taken the roller to the next level with its OZMO Roller 2.0, TrueEdge 3.0 mop extension and BLAST 19,500 Pascal suction.

Australian Review: ECOVACS DEEBOT X11 OmniCyclone robovac/roller mop

As of 15/9/25 Firmware 1.47.112, this is a pre-release model. There is an X11 Pro OMNI without the Cyclone Omni base at $2699.

Website	Website Product Page Manual
Price	$2,999
From	Bing Lee, Harvey Norman, JB HiFi, Retravision, The Good Guys, and ECOVACS Online. Avoid online and international stores, as they don’t have an AU warranty.
Colours	Black
Warranty	2.5-year ACL
Made in	China
Company	Ecovacs Robotics is a Chinese technology company best known for developing in-home robotic appliances. Founded in 1998, Ecovacs has a commanding lead in the Chinese market and is #1 in Australia.
About	CyberShack cleaning news and reviews CyberShack Ecovacs news and reviews

Ratings

We use the following ratings for many of the items below. CyberShack regards a score between 70 and 80/100 as a fit-for-purpose pass mark. You can click on most images to enlarge them.

• Fail (below expectations), and we will let you know if this affects its use.
• Pass(able) rating that is not as good as it should be.
• Pass (meets expectations).
• Pass ‘+’ rating to show it is good, but does not quite make it to Exceed
• Exceed (surpasses expectations or is the class leader).

First Impression – It is getting harder to up the ante

I have been reviewing robot vacuum/mops for years. It got increasingly difficult to work out which were good, better and best, so we wrote the aforementioned guide, which again needs an update (soon).

So, I won’t enter into a debate yet about which robot roller mop does the best job, because from what I can see, they are very much better than the 2024 rotating mop pads. And frankly, they all cost pretty much the same. So, wait for event-related sales like Black Friday to get 30-50% off!

What I will say is that in these Gen 6 AI-bots, we all want:

• Unattended, whole-of-home cleaning ✅
• Better obstacle clean and closer avoidance ✅
• Pet programs ✅
• AI to better clean different floor types, including carpet ✅
• Better edge and corner clean on hard floors ✅
• Higher sill climbing ✅
• More suction to remove more dirt from carpet ✅
• More mop lift over carpet ✅
• More mopping power for hard floors ✅
• Longer battery life and fast charge ✅
• Better base station cleaning, including hot water mop wash and hot air dry ✅
• Security and firmware updates for at least three years ✅

We could go on, but suffice it to say the ECOVACS DEEBOT OmniCyclone robovac/roller mop meets these requirements and more.

Is it better or worse than the Eufy Omni S2: Robovac with a roller mop exceeds clean (review). Both have their strengths, and you would be happy with either.

Size – typical round robot

It is approx. 35cm round, and 98 mm high. It’s a little high to get under most cupboard overhangs and uses the front bumper as a height indicator.

This is the same height and 4cm wider than the square-shaped Eufy S2. Our panellists all commented that the Eufy got into tighter spaces.

Sensors – Gen 5

It does not have a top 360° LiDAR turret. Navigation is solely by dToF and AI. As a result, it can only produce a 2D map.

• dToF Navigation (2D, 120° range to about 10m)
• AIVI 3D 3.0 RGBD front camera powered by the Visual-Language Model for object recognition and contour.
• LED light
• Right side IR edge sensor
• Carpet sensor
• Cliff sensor
• 180° Bumper

It uses a Rockchip RK3562 4-core, 4-thread, 2 GHz SoC with a built-in NPU (for AI), Mali-G52 2D GPU (for AI image processing). This processor is usually used in entry-level Android Tablets and is comparable to the entry-level Qualcomm SD4 series. It is more powerful than typical robovac processors.

AI

Where other brands are clear on the extent of AI, Ecovacs has not revealed what AI features are part of the operating system. We deduce:

• Its voice assistant, now called Agent Yiko, is more aware of commands. It is described as a self-evolving autonomous AI agent.
• AIVI 3D 3.0 object detection is largely AI-driven. There are no claims of the number of objects or stains recognised, etc.
• dToF does not use structured light or other means to navigate, so we can only assume it has AI assistance.
• AI Strategic cleaning identifies stains and how best to clean them.
• Agent Hosting Mode is not defined, but it seems to be about getting to know the layout and more efficient cleaning.

App

There are 14 pages and even more sub-pages that, while giving total control and confused the hell out of our erudite panellists. Even AI Agent Hosting was not obvious and just appeared after a few cleans. Ecovacs could well reduce the options when Agent Hosting is enabled.

Quick map – Pass+

You run a quick map on the first use or to update the maps. It is fast. It only named one room (below average) and took 10 minutes per 100m2 (about average)

We set a no-go zone, and it stopped about 30cm short of the zone, leaving an uncleaned area. We adjusted the zone slightly to compensate, but no matter how minute the adjustment, it was either 30 cm short or 10cm too long. We have never had this issue, so we set a physical barrier.

Navigation: Pass

Because it does not have a 360° LiDAR turret, it uses 120° 2D dToF (Direct Time-of-Flight), which sends out nanosecond pulsed laser light and measures the time it takes to return. It is accurate out to about 10m.

Technically, it is not as accurate as 3D dToF, and LiDAR LDS combined and can only produce a 2D map. As it cannot do 3D, it misses the nuances, such as overhead curtains, furniture height, and overall spatial awareness. That explains the lack of room and furniture naming prowess.

It always found its way, but often spent time establishing location. On several occasions, it would cross the floor and return to the other side to continue its job.

Obstacle and Torture test: Pass

It uses the AIVI 3D 3.0 camera and AI to recognise obstacles. Ecovacs does not specify the number or types of obstacles it recognises; it only says it was AI-trained based on billions of articles.

While it did recognise all physical obstacles, including a 40×20 mm Lego block, it left quite a wide berth up to 10 cm. The app has settings for high sensitivity (default off) and Protective Navigation mode (default off), which, when enabled, take an even wider berth around obstacles.

It can recognise and navigate around pet poo. It needs to be semi-chunky rather than ‘poo’rrhea.

While it recognised some opaque or coloured liquids, it did not recognise clear liquid spills and ploughed through. See the mopping test later.

We have a blind alley torture test that has only one way in and the same way out, with limited room for the robot to manoeuvre its way out. It approached the area but failed by ignoring it completely.

The 98mm height and no turret mean it can get under most cupboard overhangs, but the right side sensor sees the cupboard door and so can’t get closer.

Vacuum: Pass on carpet and Exceed on hard floor

It has a 14cm single rotating rubber and bristle brush. It does not appear to ‘beat’ the floor like some others.

Our test is with 100g/ml (25g each) of sand, rice, oats, and Kellogg’s Nutrigrain. We do two runs and measure the weight of the collected detritus after each. Figures are 1/2 passes.

• Short pile carpet to 6mm: 55/65% with difficulty with sand and static-charged detritus.
• Medium Pile to 10mm: 62/67%.
• Long pile to 15mm: 27% and not much more on the second run. While the mop lifts 10mm, it is insufficient for a plush-cut long pile.
• Shag pile or feature rugs to 25mm: No
• Hardwood (vacuum and mop): 92/95% first run

The ZeroTangle 3.0 roller brush did a good job of clearing hair.

The right whisker does not extend as such, but it is asymmetrical and arcs out wider than the robot. It did not flick larger Nutrigrain out of the way like many robovacs. But we noticed considerable wear, and this may need to be replaced frequently.

Mopping Hard floor:  Exceed

The MagiScour 170 x 75mm mop has reinforced nylon bristles and does a better job than its predecessor.

The mop lifts 10mm over the carpet, and the auto-detection was accurate. However, 10mm is not a lot, and it wetted some of the panellists’ medium pile carpet. There are settings to vacuum the carpet first (with a dry mop), which may fix that. It also got stuck in a longer carpet, dragging its undercarriage through the pile.

In some cases, the mop extends to clean closer, but in most, it leaves a 20mm gap, and more around the obstacle. It is being cautious and could use a more aggressive setting.

Given the roller mop’s inherent efficiency, it presented a cleaner, streak-free floor than other mop types. We could not test the cleaning solution as none was provided.

Liquid: Passable

Unlike some brands (Dreame X50 Ultra, Eufy S2), when it hits a puddle of clear liquid, it does not stop suction, or the rotating brush or the side whisker (nor lift them), meaning liquid gets into the brush, vacuum throat, detritus bin, etc. This is not pet-friendly.

It detected most opaque or coloured liquids and lifted the main roller brush and retracted the side whisker. We hope a firmware update will enable AI to detect all liquid messes.

Hard floor Edge and Corner clean: Pass

The TrueEdge 3D sensor works well on standard 75-100mm kickboards. The mop extends past the robot, and two rubber wheels ensure it runs true.

When it encounters overhanging cupboard doors, however, it thinks these are the wall and won’t go under the overhang. However, it cleans within 1cm of the doors.

Carpet edge and corner clean: Fail

The mop retracts over carpet, leaving the 14cm ZeroTangle rotating brush to clean. The robot is 350mm wide, leaving about 110mm on either side that can’t be vacuumed. All robots have this issue.

The whisker did not corner clean on the carpet.

Sill: Pass+

 It has TruePass Adaptive 4 Wheel Drive climbing. In reality, the standard tractor wheels are aided by hard, lifting levers. It can get over 24mm sill and 40mm two-part sills.

Clean time: Pass

In all, we had 14 runs with different floor types, and we tested on standard and Agent Hosting AI modes.

Our 50m2 test area took

• Standard: 75 minutes, used 50% battery and about 33% of the clean water. 3 returns to the dock
• Agent Hosting: 83 minutes, 58% battery and about 38% of the clean water. 4 returns to the dock.

That is relatively slow at 1.5 to 1.66m2 a minute, but we would rather have clean than fast.

Now with other robovacs, AI cleaning usually results in less time as it gets to know the home and works out the best clean. All panellists noted that AI cleaning took longer, and the robovac returned to several areas it had previously cleaned. Again, we suspect early firmware.

Carpet vacuum will take about twice as long as it would to do two passes to cover the 14cm rotary bush. Similarly, mopping takes twice as long as a unit with a 30cm roller.

Coverage: Pass

We found that 150m2 of hard floor depletes the clean water tank. Ecovacs has a new fast GaN charge that can top up the battery by up to 6% when it comes in for a mop clean. Yes, we saw that in action and have taken this into account in the battery readings. 

Pets: Not sure

Some robots have dedicated pet settings (like Dreame X50 Ultra and Eufy S2) that won the hearts of our pet-loving panellists.

We could not find any such settings, although this may be part of the AI Agent option that works out how best to clean each area.

The tangle-free rotating brush works well on long pet and human hair.

Battery: PowerBoost is a nice idea, but mostly hype

It has a 14.4V/6.4A/92W Li-Ion pouch-type battery (ECOVACS S19-LI-144-6400). It is the same as used in earlier X models, and you can get a DIY Battery Replacement AU for approx. $130.

PowerBoost refers to a 16.4/9.5A/155.8W GaN fast charge for a few minutes while the robot docks for a quick mop wash. It claims 3 minutes for a 6% top-up that theoretically enables an ‘unlimited runtime’. It cannot sustain longer periods of PowerBoost.

Otherwise, it charges at 16.4V/3A/49W, taking about between 3 and 4 hours.

The OmniCyclone dock takes:

• Charging: 230V/2.5A/575W (3-4 hours)
• Emptying: 230V/4.5A/1035W (few minutes)
• Hot Water Washing Mop:12.5A/2875W (few minutes)
• Mop Dry: 230V/6.5A/1495W (adjustable 2, 3, or 4 hours).

With power now at 35+ cents per kWh (shoulder rates), it could cost a dollar or so each use. There is an off-peak charging option.

Robot Internal Capacities: Pass

• Clean water 240ml
• Wastewater 220ml
• Dustbin 250 ml

These are a little smaller than most, but are more than adequate if you set the correct times to return to the dock. You will get notifications about whether these were empty or full. It appears that the dustbin does not have a sensor.

Noise: Pass

<65dB except for station cleaning.

Live View Settings: Pass

It’s a low-res, fixed focus, 480 x 360 RGBD (Red-Green-Blue with depth) primarily for obstacle recognition. But you can set it to use a on-screen joystick or patrol mode. You can record video and use your phone to hear and speak.

Ecovacs states that the video remains on the device and is not a privacy concern.

Build quality: Exceed

The ECOVACS DEEBOT X11 OmniCyclone is a premium product with a premium build and a 2.5-year warranty. It looks well-built.

OmniCyclone station: Exceed

It’s a big station at nearly 400 (W) x 500 (H) x 500 (D). It needs the obligatory 1.5 metres clear in front and .5m each side. Power is from the rear.

The design feature is the PureCyclone 2.0 Auto Empty technology, which is a lift-out bagless dustbin that holds 1.6 litres of detritus. The bin is easily disassembled for cleaning under water.

One thing we noticed over a few weeks was the buildup of dust inside the bin that takes the shine off its visage. A quick wash restores that. But every cyclone bin we have seen from Dyson to the cheapies eventually gets pitted and marked,

It has a 3.2L clean water tank and claims a 400m2 mopping area, which is more like 150 m2 on default settings for hard floors.

The 2.7L wastewater tank is more than adequate, but the 220ml robot internal wastewater tank means more frequent returns to the dock.

Both tanks have water level detection.

Mop cleaning and drying

It dries at 75°C using a soaking function to help remove stains. It is then dried with hot air at 63°C for 2, 3 or 4h.

Cleaning solution

It has two tanks – one for the standard floor clean and a heavy-duty tank – neither was supplied and could not be tested. As far as we can tell, the heavy-duty solution is only for mop clean (if installed).

Maintenance: Pass+

It self-cleans as it mops and does a more intensive clean at the dock. At worst, you have to wash the internal cleaning deck occasionally.

You can remove the roller and clean around it, but the motor does not lift up, so that is a pretty fruitless task.

There are no spare parts or accessory prices on the AU website. The 1L cleaning solution is $37.50 plus freight. Ecovacs usually has reasonable costs and offers bundles.

Part	Maintenance Frequency	Replacement Frequency
Washable Roller Mop	/	Every 1-2 months
Dirty Water Box	Every week	/
Drip Tray	Every week	/
Side Brush	Every 2 weeks	Every 3-6 months. $42 per pair
Main Brush	Every week	Every 6-12 months $39.90 for two.
Filter	Every week	Every 3-6 months $14.90 for two.
Universal Wheel Anti-drop Sensors Bumper Charging Contacts on DEEBOT Charging Contacts on OMNI Station Navigation Module Navigation Module AIVI Camera	Approx every month	/
Clean Water Tank	Every 3 months	/
Dirty Water Tank	Every month	/
Body and Bottom of the  Station	Every month	/
Mop Washing Tray	Every 5 months	/
Cleaning Solution Unit	Every month	/
Dust Container	After each use	/
Dust Container Filter	Every 3-6 months	Every 3-6 months/after multiple washes

CyberShack’s view: ECOVACS DEEBOT X11 OmniCyclone robovac/roller mop does a great job – but it could be even better.

The panel and I want to make it clear that this is an excellent robot worthy of your consideration for a premium roller mop. Please take our comments more as what we think/expect the X11 should be. This could be even better with some firmware upgrades rather than any underlying hardware or design issues.

There are better Gen 5 robots in specific areas:

• dToF and LiDAR LDS combination for 3D mapping and spatial awareness. 2D is limiting.
• Object recognition. While AI does not need to know it’s a ‘cat’, it does not hurt to know more than its spatial outline.
• AI cleaning takes longer and is perhaps unnecessarily repetitive.
• Higher Sill climbing. 4WD 24mm is good but not class-leading.
• Better carpet vacuum. 19000 Pa, even with BLAST, is not class-leading.
• Mop lift of 10mm is far below the class-leader.
• Effective pet programs (this should be easy)
• Vacuum off, and rotary and whisker stop over clear liquids
• Ozonated water instead of or in addition to the cleaning solution
• The 17cm roller does not measure up to the 30cm class leader, which means more passes are required to clean the same area.

But none are deal breakers – it will work for you.

Panel final Comments

All hard floors over one level: The vacuum did a great job (on hard floors). Mopping was also excellent, but dried milk or coffee stains are still challenging. Its obstacle avoidance and edge cleaning are good, but need improvement to be class-leading.

Mix hard and carpet floors on one level: The hard floor vacuum was excellent, and the carpets were decidedly average, requiring at least two recleans. Mopping was efficient but did not remove dried stains. No home prep is needed.

Pet owner: The rotating brush and roller mop did not get clogged with pet hair. It avoided our faux test pet poo (50 x 25mm). Regarding clear liquids, it ploughs through, making an unholy mess in the vacuum dustbin.

Two-level—tiles on L0 and carpet L1 (retired Engineer): I have no complaints about the vacuum/mop on tiles. Multi-floor mapping works, but you must move the base station from floor to floor. The mop wets a medium-height carpet. Sill climbing is OK.

30/10/60% carpet/tiles/bamboo hardwood single level (reviewer—me). It did everything expected of a Gen 5 robot. Battery life on defaults and AI Intelligence is around 150 minutes. Edge and corner clean are good on hard floors and fail on carpet. The carpet vacuum was decidedly average.

Panellist’s summary:

• Strength: Mopping. The barefoot test feels squeaky clean, and it eventually removes dried-on stains.
• Above Average: Nothing really – it does the job as advertised
• Hard floor Edge clean: The mop is suitable for cleaning up to about 1cm from the wall
• Obstacle avoidance: Average and needs clear liquid recognition
• AI: Taking over the world, yet the app is incredibly detailed and beyond most. Let AI decide.
• Criticism: While it is a very good device, the OmniCyclone is not enough of a reason to buy.
• My take: The X8 was good and addressed a lot of the issues that its competitor, the Eufy S1 Pro, had. The Eufy S2 is a quantum leap over the S1 and presents new features that the X11 has not included – yet.

Ecovacs Deebot X11 OmniCyclone Ratings

It is a Gen 5 BrainyBot that can do one-pass, whole-of-home, unattended cleaning with little to no home preparation. It is not quite at Gen 6 AI-bot standard, but I am sure Ecovacs techs are working on firmware upgrades to get it well over the line.

• Features: 90. It has every Gen 5 feature plus a roller extendable mopping, decent obstacle avoidance and the OmniCyclone.
• Value: 85. It is in the ballpark, but really a good candidate for event-related sales.
• Performance: 85. It is 10/10 for mopping, but overall, it does a similar job on carpets to other premium Gen 5 robots.
• Ease of Use: 85. It is Easy to set up, and the quick map is excellent. Loses points on things that could be improved by firmware.
• Design: 85. Well-made, but robots are getting smaller and more agile and base stations are less monolithic.