What “household chores” means on this page
This URL is the buying guide for everyday housework that is not laundry and not cooking. Laundry has its own page — what a humanoid can actually do with laundry — because folding, hanging and delicates are a different skill with a different failure mode. Cooking and skill-packs live on the skill-packs guide and the kitchen hub. Mixing those jobs into this article is how two URLs start ranking for the same query and both lose.
What remains is the work most people actually picture: tidying surfaces, carrying things from room to room, loading a dishwasher, running a vacuum, taking out trash, resetting a room after dinner. In 2026 a consumer humanoid can do a useful slice of that under supervision. It cannot do it the way a person does it, and it cannot do it the way a $400 robot vacuum does floors. The purchase only works if you are buying a supervised generalist, not a replacement for either.
What a humanoid can actually do
Carry and reset. Glasses to the dishwasher, parcels from the door, cushions back on the sofa, a blanket folded in the crude way these machines fold. This is the highest-reliability slice of housework for 2026 humanoids, because the objects are known, the routes are short, and a failure is a mess rather than a flood.
Load a dishwasher. Not wash a sink full of yesterday’s pans, and not unload into the right cupboard on the first try. Load. If plates are scraped and stacked in a consistent place, several consumer platforms can get them into the rack with supervision nearby. If the sink is chaos, you are doing the job yourself and calling it training.
Vacuum under supervision. A humanoid running a stick vac around a mapped living room is real, and it is also strictly worse than a dedicated robot vacuum on coverage, edge pickup and the ability to run while you are at work. The case for it is “I already own the humanoid and I am not buying a second floor robot,” not “this beats a Roomba.” For actual floor-cleaning as a primary job, buy the vacuum. Use the humanoid for the bits a disc cannot do: moving a chair, picking up the thing the vacuum would eat, resetting the room afterwards.
Trash and clutter. Bags to the bin, recycling to the door, toys into a basket. These are fetch-and-carry jobs with a chores label, and they are within scope. They still need a mapped home and a charging location that is not a trip hazard — the robot-ready home audit and the mapping guide are setup, not optional reading.
What it must not be asked to do
Laundry end-to-end. Transfer and simple sorting are the easy wins; folding, hanging, ironing and delicates are where 2026 platforms stall. That argument is documented on the laundry URL. Do not expect this guide’s shortlist to change it.
Unsupervised whole-home cleaning. A humanoid left to “do the house” while you are at work will dock, get stuck, or invent a mess. Dedicated floor robots can run empty-home. Humanoids in 2026 still want a person in earshot for the first months, and often after that.
Anything wet and unstructured. Mopping a sticky kitchen, scrubbing a bath, dealing with a spill that has already soaked in. Water plus uncertainty is how you get a damaged floor and a robot with a wet foot. Keep a mop. Keep a person for the wet work.
| Chore | Humanoid in 2026 | Better tool |
|---|---|---|
| Carry / reset a room | Primary, supervised | A person, if speed matters |
| Load dishwasher | Useful if plates are pre-scraped | A person, or a dishwasher with a better rack than your patience |
| Vacuum floors | Possible, worse coverage | A dedicated robot vacuum |
| Mop / wet messes | Avoid | A mop, a wet vac, a person |
| Laundry | Not this URL | The laundry guide — then a person for folding |
| Cooking | Not this URL | Skill-packs guide / kitchen hub |
| Empty-home whole-house clean | Not realistic | Robot vacuum + a cleaner |
Which robots are worth considering
For this job capability and reliability dominate. Privacy still matters — a chores robot has cameras in the kitchen — but a presence-first machine that cannot carry will not earn its keep here. The consumer cluster that can actually do housework sits at $16,000–$20,000. Platforms at $40,000–$90,000 that show up in “best for chores” roundups are commercial or research machines, and the hub no longer lists them as household picks.
| Robot | Price | HOSI | Capability | Reliability | Apartment |
|---|---|---|---|---|---|
| Figure 03 | $20,000 | 9.2 | 9.4 | 9.1 | 9.5 |
| 1X Neo | $20,000 | 9.0 | 8.9 | 8.7 | 9.6 |
| Tesla Optimus Gen 3 | $20,000 | 8.5 | 9.0 | 8.2 | 8.7 |
| Unitree G1 | $16,000 | 7.7 | 8.0 | 7.6 | 7.7 |
Figure 03 is the default for chores. Capability 9.4 and reliability 9.1 are the highest in the consumer tier, which is what you want when the robot is picking up objects that can break. Apartment compatibility 9.5 means it is not only a big-house machine.
1X Neo is the pick when the home is small or quiet, or when the same robot has to be tolerable as a presence device between chores. Apartment 9.6 and privacy 9.2 lead the category. You give up a little capability versus Figure; you gain the hardware camera cut-off and on-device processing.
Tesla Optimus Gen 3 is the other $20,000 capable option (capability 9.0). Reliability 8.2 and privacy 8.0 are the trade. Unitree G1 at $16,000 is the value door: enough capability (8.0) to be in the conversation, not enough reliability (7.6) to be the default. Presence-first robots at 6.5–6.6 capability — Sunday Companion, Neura MIPA — are the wrong hardware for this job; they belong in companionship, not in a chores shortlist.
What it costs, compared with paying for cleaning
A $20,000 robot is about $32,300 over five years, or $538 a month. The five-year breakdown is the model; financing only changes when the cash leaves. A cleaner in most US metros is $30-ish an hour. $538 buys roughly 18 hours of that — about four hours a week — and the cleaner will do a better job on the specific tasks you hired them for, with no mapping, no firmware and no supervision.
So the arithmetic that works is not “replace the cleaner.” It is “cover the hours nobody is being paid for”: the ten-minute resets after dinner, the dishwasher load before bed, the thing carried upstairs so it is not still on the hall floor in the morning. If those fragments are the pain, a generalist robot can be worth $538. If the pain is “the floors need doing twice a week to a standard,” buy a robot vacuum and a cleaner. Add a humanoid later if you still want the generalist.
Two costs people skip: a spare battery if you want evening chores after a daytime job, and the family member who becomes the administrator. Insurance is usually a conversation with the home insurer, not a new policy.
What “supervised” actually means
It does not mean standing next to the robot with a remote. It means a person is in the house, can hear a collision or a stuck loop, and can stop the routine before a glass hits the floor. The first two weeks that person is doing real work: moving a chair the map did not include, scraping a plate the robot will not, cancelling a path that heads for a cable. If that sounds like you have hired a very expensive intern, that is the right picture. The intern gets cheaper as the map improves. It does not become a cleaner.
Training time is the cost nobody puts in the spreadsheet. Plan on twenty to forty minutes a day for the first fortnight around that single chore, then a drop to “only when it yells.” Households that skip this phase and go straight to “do the downstairs” produce a robot that is unplugged by week three, which is how you get the Reddit thread that says humanoids cannot do chores. They can. They cannot be dumped on a dirty house and left.
Children and pets change the supervision rule, not the chore list. A stick vac in a toddler’s path is an injury question, which belongs on the kid and pet safety guide, not here. The chores implication is simpler: do not schedule floor work as background noise around a small child, and do not treat the robot as a babysitter that happens to vacuum.
How to set it up so it is still running in month six
Pick one chore for two weeks. Dishwasher load, or living-room reset, or trash. Not five. The first weeks are for discovering the house’s actual bottlenecks — the chair that blocks a turn, the rug that eats a foot, the cupboard the robot cannot reach — and for teaching the household not to leave a wet pan in the “robot’s pile.”
Do not skip mapping, and do not skip no-go zones on stairs you do not want it on and rooms you do not want filmed. A chores robot in the bathroom is a privacy incident, not a cleaning strategy. Keep a person in the house for unsupervised-adjacent work until the failure modes are boring. Then, and only then, widen the routine.
Who this is actually for
A household that wants a supervised generalist for carry, reset and a slice of kitchen clearing, and that already understands a robot vacuum still owns the floors. Not someone hoping to fire a cleaner. Not someone whose actual pain is laundry (read that guide) or cooking (read skill-packs). Not someone shopping an $80,000 warehouse platform because a roundup put it under “house chores.”
The house-chores hub now recommends the same four consumer robots this guide does. If you are still choosing a job rather than a robot, the Robot Finder is the faster path.
