SwitchBot
SwitchBot Onero H1
- Autonomy
- IIAssisted
- Pre-release
- 54/100Plausible
- Readiness
- Not assessed
- Price
- $9,999list
ComparisonSwitchBot Onero H1 and Unitree G1
Unitree
Unitree G1
- Autonomy
- IIAssisted
- Pre-release
- 75/100Promising
- Readiness
- Known Issues
- Price
- $13,500list
Up to three
Autonomy Ladder
- IManual
- IIAssisted
- IIIConditional
- IVEnvironmental
- VGeneralized
All humanoid robots. Every figure below is placed against the whole class, so a gap can be read against how much the class itself varies.
Where each one suits you.
The conditions each robot was assessed against, side by side. Rows marked in gold are where these machines genuinely differ.
| Condition | SwitchBot Onero H1 | Unitree G1 |
|---|---|---|
| Where it operates | ||
| Home / residentialthey differ | GoodDesigned for home use; wheeled base suited to flat-surface domestic environments with standard-height appliances. | PoorG1 requires continuous programming and operator expertise unsupported in consumer settings |
| Office / retail | DependsTask set is household-focused; commercial deployments with higher throughput or structural variability are outside the stated design domain. | DependsSuitable for pre-mapped facilities with dedicated technical staff; structured inspection tasks only |
| Warehouse / factorythey differ | PoorNo published payload rating, safety certification, or industrial protocol support; not designed or marketed for industrial environments. | GoodResearch institutions and engineering labs with embedded systems expertise can deploy and customize |
| Primary tasks | ||
| Household choresthey differ | GoodCES demonstrations covered laundry, dishwashing, window cleaning, and object manipulation as core task targets. | PoorG1 lacks task generalization; household chores require adaptive decision-making and real-time replanning |
| Social / carethey differ | DependsNo documented social-interaction or companion-mode features; the H1 is task-oriented, not conversational or presence-focused. | PoorNo natural language understanding or safe human interaction guarantees; not designed for social roles |
| Specialized jobthey differ | DependsOmniSense VLA is designed for generalization, not narrow specialization; purpose-built platforms would outperform it on single specialized tasks. | GoodExcels at pre-programmed specialized tasks in controlled, pre-mapped environments |
| Supervision needs | ||
| Fully autonomous | PoorNo shipping evidence supports unattended autonomous operation; beta status means edge-case behaviors remain uncharacterized. | PoorG1 requires continuous operator supervision for all task initiation, monitoring, and error recovery |
| With oversight | GoodBeta program and CES demonstration context assume an operator is present; supervised deployment matches current capability stage. | GoodDesigned for operator oversight; firmware includes teleoperation and manual task control |
| Teleoperated | GoodSwitchBot's mid-2026 beta is structured around pilot participants providing feedback; pilot deployment is the explicit intended use at this stage. | GoodIdeal for pilot deployments with dedicated technical teams managing customization and oversight |
| Timeline | ||
| Ready today | PoorNot shipping as of June 2026; no consumer units available and general availability is expected toward year-end 2026 at earliest. | PoorNot production-ready for consumer or general commercial deployment; research platform only |
| Near-term (1–2 yrs) | DependsEnd-of-2026 or early-2027 consumer availability is plausible if beta progresses on schedule; pricing and final task capability remain unconfirmed. | DependsUnitree's roadmap indicates capability improvements; consumer viability on multi-year horizon |
| Research phase | GoodOn-device VLA architecture and 22-DoF arm system make the H1 a substantive evaluation platform for embodied AI research during the beta period. | GoodProven in research institutions; platform actively used for locomotion and manipulation studies |
A robot being poor for a condition is a fact about fit, not a fault. “Depends” means the record names a limit worth reading, not that the answer is unknown.
How much the claim can be trusted.
None of these has shipped, so they are assessed on track record, engineering plausibility and how the demonstrations match the claims.
Track Record35% of score
Has this maker shipped what it demoed before, on schedule, with the promised capability?
4885
- SwitchBot Onero H148
- Unitree G185
28median 69 across 80 pre-release records100
Engineering35% of score
Are the demonstrated capabilities achievable in shipping form, or do they need breakthroughs not in evidence?
6174
- SwitchBot Onero H161
- Unitree G174
35median 70 across 87 pre-release records100
Demo Match30% of score
What has been shown to work, against what is claimed in marketing.
5466
- SwitchBot Onero H154
- Unitree G166
30median 66 across 87 pre-release records100
Published beside the number
Never averaged into it. These describe how much the maker has disclosed, not what the machine does.
| SwitchBot Onero H1 | Unitree G1 | |
|---|---|---|
| ReadinessShip-date specificity, pricing transparency, distribution and regulatory status. Published beside the number; never averaged in. | 38/100 | 90/100 |
| DisclosureHow much of the product is disclosed. Higher means fewer fundamental unknowns remain. Published beside the number; never averaged in. | 35/100 | 70/100 |
How far apart is a long way?
A number only means something next to the spread it sits in. Each track is the real range across every classified robot in the class, with the median marked.
Price18 with a published figure
$9,999$13,500
- SwitchBot Onero H1$9,999
- Unitree G1$13,500
$1,385median $22,500$250,000
Height51 with a published figure
51.2 in52 in
- SwitchBot Onero H151.2 in
- Unitree G152 in
34.7 inmedian 66.9 in86.6 in
Not compared, because fewer than two of these robots carry a published figure: Runtime, Charge time, Noise, Weight.
What the numbers leave out.
Written on each record, and the only thing here every robot on the site carries.
The trade-off
What each design gives up to be good at what it is good at.
SwitchBot Onero H1
The wheeled base makes flat-surface navigation tractable but rules out the stair-climbing and floor-transition capability that household generalization requires.
On-device VLA processing preserves privacy and offline function; it also means capability is bounded by what the embedded hardware can run, with no published ceiling on task complexity.
Unitree G1
Accessible platform design versus limited autonomous decision-making. Scripted architecture lowers research barriers but restricts real-world adaptability.
Dexterous manipulation versus energy constraints. Multi-DOF gripper enables complex tasks but consumes battery reserves, reducing session duration.
Who each one is wrong for
Stated by the record itself. A robot being wrong for a household is a fact about fit, not a fault.
SwitchBot Onero H1
- Households expecting autonomous, unsupervised daily chores
- No shipping evidence supports unattended, repeatable task execution. CES demonstrations were controlled; the platform is in pre-release beta. Autonomous independence across varied household conditions is not documented.
- Multi-floor homes with stairs or significant elevation changes
- The Onero H1 uses a wheeled cylindrical base with no documented stair-climbing capability. Press and manufacturer materials describe it as designed for flat-surface navigation only.
- Buyers on a budget or seeking near-term ROI
- At $9,999 pre-release price with no consumer shipping date confirmed and task speed demonstrated at roughly two minutes per laundry sequence, the platform is not positioned for cost-justified daily utility.
- Households with heavy clutter, uneven floors, or irregular layouts
- The VLA model adapts to shape and position of objects but CES demos operated in staged environments. Complex, variable floorplans introduce conditions not yet documented as within the robot's operating design domain.
- Buyers without an existing SwitchBot ecosystem
- SwitchBot's Smart Home 2.0 positions the H1 as an orchestrator of its own device lineup. Households without SwitchBot vacuums, hubs, or connected appliances lose the primary coordination layer the robot is designed around.
Unitree G1
- Homeowners and small businesses
- G1 requires continuous programming, lacks home integration standards, and demands embedded-systems support unavailable to consumers.
- Unstructured or dynamic environments
- Outdoor terrain, crowded spaces, and environments with unexpected obstacles exceed the platform's perception and decision-making capabilities.
- Unsupervised autonomous operation
- G1 cannot switch between unplanned tasks or recover from deviations without operator intervention and re-programming.
- Organizations without technical staff
- Deployment and maintenance require embedded systems expertise and ongoing firmware management.
What stays your job
The upkeep ownership hands back to you, whatever the autonomy level says.
SwitchBot Onero H1
- After every use: Return the robot to its charging dock; SwitchBot has not published runtime per charge.
- As needed: Clean the Intel RealSense camera lenses on the head, arms, and abdomen to maintain visual accuracy.
- As needed: Apply software and task-model updates through the SwitchBot app; OmniSense VLA update cadence is not specified.
- As needed: Inspect arm and hand joints for wear; SwitchBot documents no service interval.
- As needed: Expect this schedule to change; the H1 is listed as coming soon, enters beta in mid-2026, and has no published owner maintenance documentation.
Unitree G1
- Daily: Battery charge cycle, visual inspection of joint actuators and gripper for wear debris.
- Weekly: IMU and joint encoder recalibration, firmware update checks and testing.
- Monthly: Gripper servo inspection and pad replacement if worn, joint lubrication if needed.
- Every 3-6 months: Actuator firmware deep inspection, full sensor calibration cycle, structural integrity check.
- As needed: Gripper pad replacement, joint actuator service after environmental contamination or high-impact motion.
How much each of these is standing on.
Classifications rest on different depths of evidence. A pre-release assessment and a verified one are not the same kind of claim.
| Robot | Evidence status | Sources reviewed | Last assessed |
|---|---|---|---|
| SwitchBot Onero H1 | Pre-Release | 1 | Aug 2026 |
| Unitree G1 | Not Consumer-Ready | 13 | Sep 2026 |
Evidence status counts the sources behind a classification, not their kind. A high count can be maker documentation and price coverage, so read it as how much was reviewed rather than as independent testing. How we weigh evidence.
Keep going.
The full recordSwitchBot Onero H1Specifications, evidence, the Fit Check in full, and every tracked change.
The full recordUnitree G1Specifications, evidence, the Fit Check in full, and every tracked change.Robovations classifies robots; it does not rank them. Nothing on this page is a recommendation, and no manufacturer influences a classification or its placement here.
Add a robot
Comparing across classes is allowed — the page will switch to per-class positioning and tell you it has.