Find, compare and buy robots from top manufacturers

1X NEO vs Unitree H2: Which Humanoid Is Better?

1X NEO and Unitree H2 target completely different buyers. This comparison examines price, autonomy, strength, hands, AI, safety and availability to determine which humanoid is better for the home, research and development in 2026.

Image Credits:
1X Technologies & Unitree Robotics

Miguel Anton

Editor

1X NEO and Unitree H2 are two of the most advanced humanoid robots that an individual or organisation can currently order, but they are built for fundamentally different buyers.

NEO is a lightweight, soft-bodied home robot designed to perform domestic chores, interact naturally with people and improve through a combination of onboard AI, software updates and scheduled remote expert assistance. Unitree H2 is a larger, heavier and more powerful humanoid platform aimed primarily at research, demonstrations, embodied-AI development and controlled technical environments.

The short answer is clear: 1X NEO is the better choice for a consumer who wants a humanoid designed around the realities of a home. Unitree H2 is the better choice for a laboratory, university, robotics company or developer that needs a configurable full-size humanoid platform—provided it selects the H2 EDU configuration rather than assuming the $29,900 standard H2 includes development access and dexterous hands.

Neither robot is a finished universal worker. NEO is entering early consumer deliveries with basic autonomy and scheduled expert support, while Unitree explicitly warns that some H2 functions are still being developed and that users must keep a safe distance from the robot. The correct comparison is therefore not simply “Which robot has the bigger number?” It is a domestic product architecture versus a research and demonstration platform.


Quick verdict: Choose 1X NEO for home use, safer physical interaction, autonomous charging, advanced hands, natural voice interaction and a consumer purchase or subscription path. Choose Unitree H2 EDU for secondary development, robotics research, custom compute, motion experimentation and access to a more powerful full-size mechanical platform. Do not buy the standard H2 expecting an open development kit, and do not buy NEO expecting fully autonomous completion of every household chore from day one.


Anton Robots maintains detailed profiles for 1X NEO and Unitree H2. You can also compare 1X NEO and Unitree H2 side by side using the Anton Robots comparison tool.

Last reviewed: 17 July 2026. This comparison separates current published specifications, latest announced hardware, demonstrated capabilities, order terms and future software claims. Prices exclude any taxes, shipping, customs duties, optional hardware or regional costs unless stated otherwise.

1X NEO vs Unitree H2 at a Glance

Comparison1X NEOUnitree H2Winner in 2026
Product typeConsumer-oriented humanoid home robotFull-size humanoid research, demonstration and embodied-AI platformDifferent purposes
Current commercial statusPre-order and early production; first US deliveries scheduled to begin in 2026Listed for sale; H2 EDU available through sales enquiryH2 for conventional platform purchasing; NEO for consumer pre-order
Published starting price$20,000 Early Access ownership or $499/month subscription; $200 refundable deposit$29,900 for standard H2; H2 EDU priced by quotationNEO on headline price
Primary buyerEarly-adopter households and consumersUniversities, laboratories, robotics developers, exhibitors and technical organisationsDepends on buyer
Primary environmentIndoor domestic environmentsIndoor controlled environments with trained operators and safety separationNEO for homes; H2 for labs
Height5 ft 6 in (167.6 cm)182 cm (approximately 5 ft 11.7 in)Depends on required form factor
Weight66 lb (approximately 30 kg)Approximately 70 kg (154 lb)NEO for lower mass and home suitability
Body constructionSoft custom lattice-polymer body, knit suit and covered jointsAircraft-grade aluminium, titanium alloy and engineering plasticsNEO for passive home safety; H2 for rigid mechanical strength
Body degrees of freedomSeven per arm, six per leg, three in the neck and two in the spine, plus articulated hands31 body joints: six per leg, seven per arm, three waist and two headNot directly comparable because hand DoF are treated differently
HandsLatest announced shipping hand has 25 DoF per hand, force control and tactile sensing; another current product table still lists 22 DoF per handNo dexterous hands on standard H2; H2 EDU supports multiple optional hand modelsNEO as supplied
Arm payload18 lb (8.2 kg) manufacturer-stated arm payloadRated approximately 7 kg; peak approximately 15 kg, configuration and posture dependentH2 on peak mechanical rating; NEO on clearly integrated manipulation
Whole-body carrying capacity55 lb (approximately 25 kg)No directly equivalent whole-body carry rating publishedNEO for published carry specification
Maximum lift claim154 lb (approximately 70 kg), distinct from sustained carrying or arm payloadNo directly equivalent whole-body lift figure publishedNEO for disclosure, not a like-for-like strength win
Maximum joint torqueNo directly comparable body-joint torque table published120 N·m arm joint and 360 N·m leg joint maximumH2 for published high-power joint ratings
Walking speed1.4 m/sNot publicly listed for H2NEO for specification transparency
Maximum running speedUp to 6.2 m/s manufacturer-statedNot publicly listed for H2NEO for published figure; real operating limits still matter
Battery capacity842 Wh972 WhH2 on stored energy
Published runtimeUp to four hoursApproximately three hoursNEO
Charging workflowSelf-charging; NEO plugs itself inQuick-release smart battery and supplied charger; automatic docking not publishedNEO
Standard compute1X NEO Cortex based on NVIDIA Jetson Thor, advertised at up to 2070 FP4 TFLOPSParameter table lists Intel Core i5 for platform capability; product marketing also references 2070 TOPS, while high-power modules are configuration dependentNEO for a clearer integrated shipping compute statement
PerceptionDual 8.85 MP 90 Hz stereo fisheye cameras, microphones, tactile hands and joint-force feedbackWide-field binocular camera, microphone array and onboard platform sensingNEO for published sensor detail and touch
ConnectivityWi-Fi, Bluetooth and 5GWi-Fi 6 and Bluetooth 5.2NEO for 5G; otherwise use-case dependent
AutonomyRedwood generalist AI, basic autonomy at launch, onboard inference and growing capabilitiesNo equivalent supported general-purpose autonomous task catalogue published for standard H2NEO for productised autonomy
TeleoperationOwner remote control plus scheduled 1X Expert supervision for complex tasksManual controller included; custom teleoperation and development depend on configurationNEO for consumer workflow; H2 EDU for developer control
Secondary developmentNo open low-level public development package positioned as the primary consumer offerNot supported on standard H2; supported on H2 EDUH2 EDU
Ingress protectionIP68 hands and IP44 bodyNo complete H2 ingress-protection rating publishedNEO
Published warrantyThree-year warranty with Early Access ownershipEight months standard H2; 12 months H2 EDUNEO
Best reason to choose itPurpose-built home design, advanced hands, lower mass, consumer interaction and self-chargingFull-size mechanical platform, strong joints, configurable EDU version and development accessNEO for households; H2 EDU for technical teams

What Are We Actually Comparing?

NEO and H2 share a humanoid silhouette, but they are not substitutes sold for the same job.

A buyer comparing only price, height, payload and computing power could easily reach the wrong conclusion. NEO’s hardware, software, support and physical design are organised around putting a robot in an ordinary home. H2’s rigid body, high joint torque, controller, optional compute and EDU development route are organised around technical experimentation in controlled spaces.

That difference changes how every specification should be interpreted. Lower weight is a major advantage in a home but not necessarily in a motion-control laboratory. A soft body is central to NEO’s value proposition but may be less important than actuator power and accessible control interfaces for an H2 research team. A dexterous hand is essential for domestic chores, while a laboratory may prefer to choose or replace its own end effector.

Which 1X NEO?

This comparison uses the production-intended NEO home robot currently offered through 1X’s order page, not the earlier NEO Beta or NEO Gamma prototypes.

The current NEO is listed at 5 ft 6 in, 66 lb, with an 842 Wh battery, four-hour runtime, self-charging, Redwood AI, a built-in language interface and a $20,000 Early Access ownership option. 1X says first US deliveries begin in 2026 and that early robots will arrive with foundational autonomy that improves over time.

The hand specification requires special care. 1X’s main order page still lists 22 DoF per hand, reflecting the specification announced with the product in 2025. On 9 July 2026, however, 1X announced a new shipping hand with 25 DoF: 22 fully actuated joints in the fingers and palm plus three at the wrist. 1X states that these are the hands intended to ship on every NEO. This article therefore treats 25 DoF as the latest announced hand specification while preserving the discrepancy so buyers know to confirm the exact configuration attached to their order.

Which Unitree H2?

Unitree sells more than one H2-related product, and combining them creates misleading specifications.

The standard Unitree H2 is the $29,900 base model. It has 31 body joints, a wide-field binocular camera, an Intel Core i5 platform computer, a quick-release battery, approximately three hours of runtime, a manual controller and no dexterous hand as standard. Unitree’s table also states that the standard model does not support secondary development.

The Unitree H2 EDU is the development-oriented version. It is priced by quotation, supports secondary development, can be configured with an Intel Core i7 development computer, optional high-power computing modules and multiple dexterous-hand options, and carries a 12-month published warranty instead of the standard model’s eight months.

The Unitree H2 Plus is a separate, more expensive full-stack research product associated with NVIDIA Isaac GR00T and is listed separately by Unitree. Its compute, hands, software and $100,000 official-store price should not be silently assigned to the $29,900 H2.

The central rule of this comparison is simple: the $20,000 production-intended NEO is compared with the $29,900 standard H2, while H2 EDU, H2 Plus and optional Unitree hands are labelled as separate configurations. A video, optional module or future software feature is not treated as standard equipment.

How We Compared 1X NEO and Unitree H2

This comparison prioritises current manufacturer product pages, official order pages, official technical documentation and direct company engineering updates. It does not use reseller descriptions to fill missing specifications or assume that visually demonstrated motions are autonomous, repeatable or included in the purchased product.

Every material claim is treated as one of five types:

  • Current published specification: a number or capability shown on the present official product or order page.
  • Latest announced hardware: a newer component that the manufacturer says will ship, even when an older product table has not yet been updated.
  • Demonstrated capability: an action shown in company material without assuming universal autonomy, reliability or customer access.
  • Configuration-dependent capability: a feature available only through EDU, Plus, optional compute, optional hands or another paid configuration.
  • Not published: information that cannot responsibly be inferred from marketing language, an unrelated model or a short video.

This approach matters because humanoid specifications are unusually easy to misread. A peak payload is not a continuous payload. Battery capacity is not productive runtime. Compute throughput stated in TOPS is not directly comparable with FP4 TFLOPS without knowing precision, architecture, software utilisation and workload. A robot that can be manually controlled through a task has not necessarily mastered that task autonomously.

Which Is Better: 1X NEO or Unitree H2?

1X NEO is the better complete product for a home. Unitree H2 EDU is the better platform for a robotics development programme.

NEO integrates the elements a household robot actually needs: a lower-mass body, compliant tendon-driven actuation, covered joints, a soft exterior, sophisticated hands, voice interaction, mobile-app control, autonomous charging, onboard AI and a consumer support model. Its hardware is designed around contact with ordinary objects and proximity to people rather than around maximum actuator output.

H2 provides a different kind of value. It is taller, more than twice as heavy, publishes substantially higher joint torque and offers a development path through H2 EDU. A laboratory can use it to study locomotion, motion imitation, manipulation, embodied models, custom compute and human-scale robot control. It is much closer to a research instrument than a domestic appliance.

The standard H2 complicates the decision. At $29,900 it is less expensive than many full-size humanoids, but Unitree’s own table says it lacks secondary development and a dexterous hand. That means the base model may be attractive for demonstrations, display, supported motions and platform evaluation, yet the buyer that wants to build new applications should request an H2 EDU quote before comparing budgets.

NEO also has an important limitation: 1X does not claim that every early robot will arrive as a fully autonomous housekeeper. Its order page says early owners receive basic autonomy and that 1X Experts can remotely supervise complex tasks at scheduled times. Buyers are purchasing an evolving product and service relationship, not a guaranteed replacement for all domestic labour.

The winner depends on the job, but the wrong pairing is easy to identify: H2 should not be selected as an unsupervised home assistant, and NEO should not be selected as an open laboratory humanoid without confirming development access.

Price and Availability

How much does 1X NEO cost?

1X lists two commercial paths:

  • Early Access ownership: $20,000, with a three-year warranty, premium support and priority delivery.
  • Standard subscription: $499 per month, with a starter productivity package and standard delivery.

A fully refundable $200 deposit is due when placing the current order. US deliveries are scheduled to begin in 2026, and 1X says some customers will receive NEO in 2026 while others will receive it later as production and product validation scale.

The ownership and subscription options are not automatically equivalent. A simple multiplication of monthly fees does not capture hardware ownership, replacement, support, included capabilities, minimum term, cancellation, regional availability or what happens at the end of a subscription. Buyers should request the complete subscription agreement before comparing lifetime cost.

The $20,000 amount also may not be the final delivered cost outside the advertised US programme. Taxes, regional delivery, service availability and future international rollout can change the amount and timing.

How much does Unitree H2 cost?

Unitree lists the standard H2 at $29,900 before tax and shipping. Customs duties, import clearance, local taxes, freight, insurance and any distributor margin can materially increase the landed cost.

H2 EDU is priced by quotation. This matters because EDU is the configuration that supports secondary development and optional dexterous hands. A technical buyer should not compare the $29,900 base H2 with NEO and then assume that development compute, hands and SDK access are included.

Unitree also lists H2 Plus separately at a substantially higher price. It is not a free software upgrade to the standard H2 and should be budgeted as a different product.

Which robot is cheaper?

NEO has the lower published ownership price by $9,900 and also offers a monthly subscription. However, it is a pre-order consumer product, while H2 is a configurable robotics platform. The lower number does not make NEO the cheaper route for a laboratory if the team cannot access the required low-level interfaces, and H2’s $29,900 price does not make it a better value if the buyer actually needs the more expensive EDU configuration, hands and compute.

Which robot is available sooner?

H2 has the clearer conventional product listing, but Unitree does not publish a universal in-stock delivery time for every region and configuration. NEO is explicitly in an early delivery programme, with first US deliveries beginning in 2026 and broader market expansion planned later.

Availability verdict: H2 is the more conventional platform purchase. NEO has the clearer consumer reservation programme but greater timing uncertainty for individual customers.

Home Robot vs Research Platform

The most important question is not which robot looks more advanced. It is whether the robot’s complete design matches the environment.

What is 1X NEO designed to do?

NEO is designed to assist with chores and everyday interaction in a home. 1X presents examples such as retrieving objects, opening doors, handling laundry, tidying, moving through rooms, communicating by voice and manipulating common household items.

Its architecture reflects that mission:

  • Low-inertia tendon drives rather than highly geared rigid actuation
  • A soft body and fully covered joints
  • Advanced tactile hands for varied household objects
  • Voice and mobile interfaces for non-technical users
  • Autonomous charging rather than manual battery management
  • Onboard AI designed for mobile bimanual manipulation
  • Remote expert assistance when autonomy is insufficient
  • A consumer ownership, warranty and subscription model

This does not prove that NEO can reliably perform every advertised chore in every home. Domestic environments contain stairs, pets, children, reflective surfaces, loose rugs, clutter, liquids, breakable objects and countless edge cases. Early buyers should evaluate exact supported tasks rather than extrapolate from the form factor.

What is Unitree H2 designed to do?

Unitree positions H2 around dynamic motion, embodied models, humanoid research and multiple work scenarios. The platform’s 31 body joints, strong actuators, rigid construction, controller and development configuration make it suitable for:

  • Humanoid locomotion and balance research
  • Motion imitation and whole-body control
  • Embodied-AI model testing
  • Teleoperation research
  • Manipulation experiments with optional hands
  • University teaching and robotics demonstrations
  • Trade-show, exhibition and entertainment applications
  • Custom perception and compute integration on H2 EDU

Unitree does not present standard H2 as a finished autonomous domestic worker. Its own product warnings advise maintaining a sufficient distance from people, note that some sample functions remain under development and encourage individual buyers to understand the limitations of current humanoids.

Use-case verdict: NEO wins for a real home product concept. H2 EDU wins for a configurable research project. Standard H2 is strongest as a platform demonstration or supported-function robot, not as a substitute for EDU development access.

Size, Weight and Physical Design

NEO is 167.6 cm tall and weighs approximately 30 kg. H2 is 182 cm tall and weighs approximately 70 kg. H2 is therefore around 14.4 cm taller and 40 kg heavier.

That weight difference is not cosmetic. A 70 kg rigid humanoid contains more energy when moving or falling and requires larger clear zones, stronger handling procedures and more cautious recovery planning. It may also be more difficult to transport, lift after a fall or move through a domestic staircase.

NEO’s lower mass is combined with a soft external structure. 1X wraps the hardware in custom 3D lattice polymer, covers access to the joints, uses a knit suit and prioritises backdrivable tendon actuation. This does not make the robot harmless, but it is a coherent passive-safety strategy for a machine intended to share living space.

H2 uses aircraft-grade aluminium, titanium alloy and high-strength engineering plastics. This structure supports high joint torque and dynamic motion, but Unitree’s published warning makes the intended operating discipline clear: users should maintain safe distance and operate with caution.

H2’s greater height can be advantageous in research involving human-scale reach, tools, work surfaces or motion capture. NEO’s shorter height and narrower domestic presentation may be easier to accommodate around furniture and doors.

Physical-design verdict: NEO is substantially better suited to a home. H2 is the stronger full-size mechanical research platform.

Payload, Strength and Useful Manipulation

Payload figures are among the easiest specifications to misuse in humanoid comparisons.

1X publishes three distinct NEO numbers:

  • Arm payload: 18 lb (8.2 kg)
  • Carry capacity: 55 lb (approximately 25 kg)
  • Lift: 154 lb (approximately 70 kg)

These figures describe different conditions. A short whole-body lift is not equivalent to carrying the same load through a home, and neither is equivalent to holding an object at full arm extension. The arm-payload figure is the most relevant starting point for manipulation, while the carry figure is more relevant to moving a load close to the body.

Unitree publishes an H2 arm maximum payload of approximately 7 kg rated and 15 kg peak. It also publishes maximum joint torque of 120 N·m for an arm joint and 360 N·m for a leg joint. Unitree notes that parameters vary with scenario and configuration, and buyers should expect payload to change with reach, pose, speed, end effector and centre of mass.

H2 has the larger peak arm-payload claim and far more published body-joint torque. NEO, however, ships around a complete manipulation system: force-controlled hands, tactile sensing and a domestic AI stack. Standard H2 does not include dexterous hands, so mechanical arm capacity alone does not tell the buyer which objects it can pick up.

Strength verdict: H2 is the more powerful mechanical platform. NEO is the more complete out-of-box manipulation proposition. For either robot, request a demonstration with the exact object, grasp, reach and repeated cycle required.

Hands and Dexterity

Hands are one of NEO’s clearest advantages.

1X’s latest hand announcement describes 25 DoF per hand: 22 fully actuated joints across the fingers and palm plus three wrist joints. The hand uses tendon drives with low gear ratios, native force control, backdrivability and tactile skin that measures pressure, contact location and shear.

1X publishes additional hand-level specifications including:

  • IP68 sealing and food-safe materials
  • Approximately ±0.2 mm positioning accuracy
  • Distal flexion forces up to 45 N
  • 17.75 N·m wrist torque
  • Slip detection through tactile and shear sensing
  • Millions of component and finger test cycles

The company has demonstrated tasks including handling screws and coins, using a screwdriver, rotating objects in-hand, zipping clothing, pouring, plugging in USB-C, wiping surfaces and handling fragile objects. These demonstrations support the breadth of the hardware, but they do not establish autonomous success rates for each task in an arbitrary customer home.

Standard H2 does not include a dexterous hand. H2 EDU supports multiple optional hand models. This gives a research buyer flexibility to select an end effector, but it increases cost, integration work and specification complexity. The total DoF, sensing, payload and software compatibility depend on the hand selected.

Hands verdict: NEO wins decisively as a complete purchased configuration. H2 EDU may be customised for specialised research, but there is no single H2 hand specification that can be compared without naming the option.

Walking, Running and Whole-Body Mobility

NEO publishes a 1.4 m/s walking speed and a maximum running speed of up to 6.2 m/s. 1X has also described a unified mobility controller capable of omnidirectional walking, sitting, standing, kneeling, getting up from the floor and using stairs with stereo vision.

These figures and demonstrations are impressive, but home buyers should focus on controlled walking, turning, threshold handling, stair policy, fall prevention and recovery rather than maximum running speed. A robot’s headline speed may be software-limited in customer use for safety and reliability.

Unitree is widely associated with dynamic humanoid motion, and H2 marketing emphasises precise dynamic movement replication and continuously updated motion algorithms. However, the current H2 product table does not publish a maximum walking or running speed. It would be inaccurate to transfer speed figures from Unitree H1, G1 or another model to H2.

H2’s 31 body joints include six per leg, seven per arm, three in the waist and two in the head. This supports expressive and complex motion, but buyers still need official operating limits for floor conditions, slopes, stairs, recovery, payload and continuous use.

Mobility verdict: NEO wins on published speed and domestic mobility features. H2 may offer stronger dynamic motion for research, but no responsible speed winner can be declared beyond published H2 data.

Perception and Sensors

NEO’s published perception package is unusually detailed for a consumer humanoid. It includes dual 8.85 MP stereo fisheye cameras operating at up to 90 Hz, four beamforming microphones, joint-level feedback, tactile hands, audio intelligence and onboard visual processing.

This combination matters because domestic manipulation cannot rely on vision alone. Transparent, deformable, reflective or partly occluded objects often require contact feedback. NEO’s force-transparent joints and tactile skin are designed to let the robot feel contact and slipping rather than infer everything from camera images.

H2 lists a humanoid wide-field binocular camera, microphone array and high-power speaker. It also supports Wi-Fi 6 and Bluetooth 5.2. The standard page does not publish the same level of camera resolution, frame rate, tactile coverage or complete sensor inventory as NEO.

A research team may add its own sensors or use configuration-specific H2 hardware, but those additions should be treated as project scope rather than standard H2 specifications.

Perception verdict: NEO has the stronger published integrated sensor package, especially for touch. H2 EDU offers more room for custom experimentation but requires a configuration-specific evaluation.

AI, Autonomy and Task Learning

NEO is the stronger productised AI proposition, but its autonomy must be described honestly.

1X’s Redwood AI is a vision-language model designed for mobile manipulation. 1X says Redwood runs onboard and combines visual input, language, proprioception and actions to perform tasks such as retrieving objects, opening doors and navigating through a home. The company also trains across teleoperated and autonomous data from its robot fleet and uses simulation and world-model research to evaluate policies.

NEO’s order page is explicit that early owners receive basic autonomy and that capabilities grow over time. For tasks NEO does not yet know, a scheduled 1X Expert can remotely supervise the robot. This hybrid model can make the product more useful before autonomy is complete, but it also means buyers should not interpret a successfully completed chore as necessarily autonomous.

Unitree states that H2 supports embodied models and continuously evolving motion algorithms through OTA updates. Its marketing page references a 2070 TOPS chip and multiple work scenarios. However, Unitree does not publish an equivalent supported autonomous task catalogue, household skill set, intervention rate or generalist policy benchmark for the standard H2.

H2’s greatest AI value is likely to emerge in EDU and Plus configurations where research teams can deploy custom models, compute and control stacks. That is a different proposition from NEO’s manufacturer-managed home autonomy.

AI verdict: NEO wins for an integrated consumer autonomy roadmap. H2 EDU wins for teams that want to build or test their own embodied-AI stack. Neither robot has public evidence that it can autonomously replace a person across unrestricted tasks.

Teleoperation and Human Assistance

Both robots can involve human control, but the operating models differ.

NEO supports owner remote control through mobile and VR interfaces. It also includes Scheduled Expert Mode, through which a 1X Expert can remotely supervise complex actions at agreed times. This is part of the product’s path from basic autonomy toward greater capability.

For a household, the advantage is practical: the robot may complete useful work before every task has been fully automated. The trade-off is that buyers must understand who can connect, when connection is permitted, what the operator can see, how recordings are handled, what network performance is required and what happens if connectivity fails.

H2 includes a manual controller. H2 EDU is the relevant version for deeper custom control, teleoperation research and secondary development. Unitree’s controller-based workflow is more appropriate for trained operators than for a household expecting appliance-like autonomy.

Teleoperation verdict: NEO offers the more complete consumer service model. H2 EDU offers the more appropriate technical control platform.

Battery Life, Charging and Continuous Operation

NEO uses an 842 Wh battery and publishes up to four hours of runtime. 1X also states that a six-minute quick charge adds approximately one hour of runtime and that NEO can plug itself in when it needs power.

H2 uses a 972 Wh quick-release smart battery and publishes approximately three hours of battery life. A charger is included, but Unitree does not publish an automatic docking workflow for H2.

The larger H2 battery stores more energy, yet the heavier robot has a shorter published runtime. That is plausible given the difference in mass, actuator power and intended use. It also illustrates why battery capacity should not be used as a proxy for efficiency or useful work.

For NEO, four hours does not mean four hours of uninterrupted maximum-payload chores. Runtime will change with locomotion, manipulation, compute, waiting, network use and charging behaviour. For H2, research teams should plan battery changes, safe shutdown, charging supervision and the effect of dynamic motions on runtime.

Battery verdict: NEO wins on published runtime and autonomous charging. H2 wins only on nominal battery capacity.

Compute: 2070 TOPS vs 2070 FP4 TFLOPS

This comparison contains a specification trap that deserves its own section.

1X lists NEO’s integrated compute as the 1X NEO Cortex based on NVIDIA Jetson Thor, with up to 2070 FP4 TFLOPS. Unitree’s H2 marketing page says the robot is powered by a 2070 TOPS chip, but its configuration table lists an Intel Core i5 as the standard platform computer, an Intel Core i7 for H2 EDU custom development and high-power modules such as Thor as optional on the EDU model.

There are two reasons not to declare a tie:

  1. The units are not necessarily equivalent. TOPS and FP4 TFLOPS can refer to different numerical formats, operations and measurement conventions.
  2. The included configuration is unclear for standard H2. Unitree’s headline and parameter table need to be reconciled in a written quote before the buyer assumes the $29,900 model includes the advertised high-performance AI module.

A buyer should request the exact chip, memory, power mode, available compute after system overhead, supported frameworks, thermal limits and whether the quoted module is included in the delivered robot.

Compute verdict: NEO has the clearer integrated consumer-compute specification. H2 EDU may be configured as a stronger research system, but the exact module and price must be confirmed.

SDK, Secondary Development and ROS Access

This category strongly favours H2 EDU.

Unitree explicitly differentiates standard H2 from H2 EDU. The standard model is marked as not supporting secondary development. H2 EDU supports secondary development, custom compute and optional dexterous hands. Unitree also provides an H2 developer documentation centre.

That means a university or robotics company should treat H2 EDU—not standard H2—as the real competitor when the project requires custom software, model deployment, sensor integration, joint control or teleoperation research.

NEO is not currently positioned as an open low-level development kit. It has owner-facing interfaces, remote control, a mobile app and a manufacturer-managed AI stack, but 1X does not present the consumer purchase as an unrestricted ROS 2 or joint-control platform.

This is not a weakness for NEO’s intended customer. Consumer products often reduce low-level access to protect safety, reliability and support. It is, however, a decisive limitation for a research team that needs control of the robot rather than access to supported household functionality.

Developer verdict: H2 EDU wins. Standard H2 does not win this category merely because it is made by a robotics platform company.

Safety Around People

Safety is the most important category in this comparison because both robots are mobile machines capable of applying significant force.

1X NEO safety design

NEO is designed around passive physical safety:

  • Approximately 30 kg total mass
  • Low-inertia tendon drives
  • High backdrivability
  • Soft custom lattice-polymer body
  • Covered joints with no external pinch points
  • Soft knit suit and shoes
  • Force-controlled, compliant hands
  • Manufacturer-stated head injury criterion below 250

This is a substantially more credible home-safety architecture than placing a rigid 70 kg research humanoid around untrained occupants. Nevertheless, manufacturer specifications are not a substitute for an independent safety case. Buyers should ask what standards, test methods, operating modes, speed limits, fall controls and child or pet policies apply to the delivered product.

Unitree H2 safety position

Unitree’s product page states that the robot has a complex structure and extremely powerful output and instructs users to keep sufficient distance from people. It also says some sample functions are still being developed and tested.

That warning should override assumptions created by polished demonstrations. H2 is not presented as a freely collaborative household appliance. A laboratory should define guarded test space, emergency-stop procedures, fall zones, remote shutdown, battery handling, operator training and recovery from a collapsed pose.

Safety winner

NEO wins decisively for intended operation near people. That does not mean NEO is risk-free or independently proven safe in every home. H2 should be treated as a powerful technical machine operated in a controlled environment.

Privacy, Cameras and Remote Access

NEO’s home-first design creates a category of questions that matters less in a conventional laboratory.

The robot includes stereo cameras, microphones, memory, remote monitoring, 5G connectivity, owner teleoperation and scheduled expert access. Those features enable assistance and learning, but they also place sensing and remote-control capability inside private spaces.

Before ordering NEO, buyers should confirm:

  • Whether video, audio, robot state or task data leave the home
  • Which data are required for core operation
  • Whether users can opt out of training-data contribution
  • How long recordings and logs are retained
  • How scheduled expert sessions are authorised and visibly indicated
  • Whether an expert can connect outside a scheduled session
  • How household members and visitors are informed
  • What happens when internet connectivity is unavailable
  • How account compromise, theft or resale is handled

H2 also contains cameras, microphones, wireless connectivity and potentially custom cloud services, but organisations can place it on managed networks and design their own data architecture. That control increases responsibility rather than eliminating privacy risk.

Privacy verdict: no automatic winner. NEO offers the more complete service, while H2 EDU gives technical organisations more potential control. The correct choice depends on written data and access terms.

Ingress Protection and Household Durability

NEO’s hands are rated IP68 and its body IP44. The new hands are described as food-safe and washable, while the soft suit and shoes are machine washable.

These details are useful in a home where the robot may encounter sinks, spills, food, dust and fabric wear. They do not make the whole robot waterproof. An IP44 body is protected against splashing, not immersion or unrestricted outdoor exposure.

Unitree does not publish a complete ingress-protection rating for H2. Buyers should therefore treat water, dust and outdoor operation as unsupported unless a written configuration document states otherwise.

Durability verdict: NEO wins on published household-oriented protection. Neither robot should be exposed to conditions beyond its documented limits.

Warranty, Maintenance and Support

1X lists a three-year warranty and premium support with Early Access ownership. Its product page also describes service replacement targets for tendon-drive components. The exact exclusions, location coverage, response time, accidental-damage policy and availability of in-home service should still be reviewed in the purchase agreement.

Unitree publishes an eight-month warranty for standard H2 and 12 months for H2 EDU, subject to its warranty documentation. International buyers must confirm where repairs occur, who pays freight, whether spare parts are available locally and how long the robot may be out of service.

Maintenance burden will also differ. NEO’s consumer model implies a more vertically integrated service experience. H2’s modular research role may allow more technical intervention, but that can require skilled staff, tools and spare assemblies.

Support verdict: NEO wins on published warranty length and consumer positioning. H2 may suit organisations capable of maintaining a technical platform, but service terms must be established before import.

Production and Delivery Scale

1X says its 58,000-square-foot Hayward factory has commenced full-scale NEO production and can produce up to 10,000 units per year when combined with current capacity plans. The company targets scaling toward more than 100,000 units annually by the end of 2027.

Those are capacity statements and plans, not proof that every reservation will be delivered immediately. 1X itself says some customers will receive NEO in 2026 and others later as it expands testing, production and product readiness.

Unitree already sells multiple robot families globally and lists H2 through its official channels. However, it does not publish H2 production output, current inventory, annual capacity or region-specific lead times on the product page.

Production verdict: 1X provides more specific NEO capacity information. Unitree provides the more established platform-sales context. Neither published figure proves the delivery date for a specific order.

Reliability and Real-World Evidence

Neither product has the kind of independent, long-duration field evidence available for mature industrial automation.

1X has published component-cycle testing, hand durability work, internal home testing and factory progress. Its latest hand announcement describes millions of component and finger cycles and wrist testing beyond two million cycles under high loads. These are meaningful engineering disclosures, but they are not equivalent to independent household uptime, successful chores per day or service rates across a large installed base.

Unitree has demonstrated impressive dynamic motion and publishes clear mechanical specifications, but it does not provide H2 customer uptime, mean time between failures, autonomous task-success rates or long-term payload-cycle data.

A buyer should request:

  • Productive uptime and intervention rate
  • Fall frequency and recovery procedure
  • Battery degradation and replacement price
  • Joint, tendon or gearbox service intervals
  • Hand and end-effector replacement cost
  • Remote diagnostic capability
  • Typical repair turnaround
  • Availability of local parts and technicians
  • Software-update rollback and support period

Reliability verdict: insufficient public field data for a winner. NEO publishes more consumer-relevant component and service information; H2 offers a platform that technical teams may be more prepared to maintain themselves.

Best Robot by Use Case

Use caseBest choiceWhy
Early-adopter home assistant1X NEODesigned for homes, lower mass, soft body, voice interface, self-charging and consumer support
Household object manipulation1X NEOIntegrated 25-DoF tactile hands and domestic AI stack
Operating near untrained people1X NEOPassive-safety architecture; H2 explicitly requires safe distance
University humanoid researchUnitree H2 EDUSecondary development, configurable compute and optional hands
Whole-body motion researchUnitree H2 EDU31 body joints, strong actuators and full-size rigid platform
Custom embodied-AI model deploymentUnitree H2 EDU or H2 PlusDevelopment-oriented configurations; exact compute and software must be specified
Consumer voice and app control1X NEOBuilt-in LLM interaction, mobile interface and scheduled expert assistance
Teleoperation researchUnitree H2 EDUBetter fit for custom development and trained technical operators
Remote household task assistance1X NEOOwner remote control and scheduled 1X Expert mode
Highest published joint torqueUnitree H2Up to 120 N·m arm and 360 N·m leg joint ratings
Strongest standard dexterous-hand package1X NEOH2 standard ships without dexterous hands
Longest published runtime1X NEOUp to four hours versus approximately three hours
Automatic charging1X NEOSelf-charging is a defined product feature
Lowest ownership price1X NEO$20,000 versus $29,900 before H2 options and import costs
Immediate fully autonomous housekeeperNeitherNEO launches with basic autonomy and expert support; H2 is not a domestic autonomy product
Open development on the base modelNeitherNEO is not positioned as an open kit; standard H2 does not support secondary development
Unsupervised use around childrenNeither should be assumedA task- and household-specific safety assessment is still required
Industrial production deploymentEvaluate purpose-built industrial robots tooNeither has a published mature industrial throughput, uptime and safety package comparable with established automation

Explore humanoid robots, home robots, educational robots and the best humanoid robots in 2026 before deciding that either platform is the right form of automation.

1X NEO Pros and Cons

Pros

  • Purpose-built for home environments rather than adapted from an industrial platform
  • Lower 30 kg mass
  • Soft lattice-polymer body and covered pinch points
  • Low-inertia, highly backdrivable tendon actuation
  • Latest announced 25-DoF force-controlled hands
  • Tactile and shear sensing for slip detection
  • IP68 hands and IP44 body
  • Published 25 kg carry capacity
  • Up to four hours of runtime
  • Automatic self-charging
  • Integrated NVIDIA Jetson Thor-based compute
  • Redwood onboard mobile-manipulation AI
  • Natural voice and mobile-app interfaces
  • Owner remote control and scheduled expert assistance
  • $20,000 ownership route and $499 monthly subscription listing
  • Three-year Early Access warranty
  • US production facility and published capacity plan

Cons

  • Early consumer product with limited independent field evidence
  • Individual delivery timing may extend beyond 2026
  • Foundational rather than universal autonomy at launch
  • Complex tasks may depend on scheduled remote expert supervision
  • Privacy and data terms are unusually important because the robot operates inside homes
  • No public low-level SDK or unrestricted ROS development positioning
  • Household task-success rates and intervention frequency are not published
  • Maximum run speed is less relevant than safe customer operating limits, which require confirmation
  • Manufacturer-stated safety metrics are not the same as independent certification for every home scenario
  • Current official pages conflict between 22-DoF and 25-DoF hand descriptions
  • International availability and support remain limited compared with the US launch
  • Subscription terms require careful review before lifetime-cost comparison

Unitree H2 Pros and Cons

Pros

  • Full-size 182 cm humanoid form
  • 31 body joints
  • High published joint torque
  • Rated approximately 7 kg and peak approximately 15 kg arm payload
  • Rigid aluminium, titanium-alloy and engineering-plastic construction
  • Wide-field binocular perception
  • Quick-release 972 Wh smart battery
  • Approximately three-hour published runtime
  • Manual controller included
  • Wi-Fi 6 and Bluetooth 5.2
  • OTA motion-algorithm updates
  • H2 EDU supports secondary development
  • H2 EDU supports optional dexterous hands and high-performance compute
  • Official starting price is low relative to many full-size humanoid research platforms
  • Strong fit for motion, locomotion and embodied-AI research

Cons

  • Standard $29,900 H2 does not support secondary development
  • Standard H2 does not include dexterous hands
  • H2 EDU pricing is not public
  • H2 Plus capabilities and price should not be attributed to the standard H2
  • Approximately 70 kg rigid body creates greater safety and handling risk
  • Unitree instructs users to maintain sufficient safe distance
  • No published automatic charging workflow
  • No complete ingress-protection rating published
  • No current maximum speed published for H2
  • No supported general-purpose autonomous task catalogue
  • Eight-month standard warranty is short for an expensive imported platform
  • Customs, shipping, service and optional hardware can materially increase total cost
  • Headline 2070 TOPS claim needs configuration confirmation against the product table
  • Some sample functions are still under development and testing
  • No public H2 uptime, intervention or long-duration workload dataset

Total Cost of Ownership

A responsible total-cost comparison must go beyond the two headline prices.

1X NEO total cost considerations

  • $20,000 ownership price or $499 monthly subscription
  • Taxes and delivery
  • Any required connectivity or service plan
  • Terms for remote expert assistance
  • Battery replacement after degradation
  • Repair exclusions and accidental damage
  • International support, if outside the initial US market
  • Software support after the warranty period
  • Whether advanced future capabilities require additional payment
  • Privacy, account and network-security management

Unitree H2 total cost considerations

  • $29,900 standard robot or higher H2 EDU quotation
  • Dexterous hands or other end effectors
  • High-performance AI compute
  • Shipping, customs duties, taxes and import brokerage
  • Safety barriers, test space and emergency systems
  • Developer labour and software integration
  • Teleoperation equipment and motion-capture systems
  • Spare batteries, chargers and replacement parts
  • Training and operator supervision
  • International return freight for repair
  • Insurance and institutional risk controls

For a household, value should be measured in useful chores completed safely with an acceptable level of intervention. For a laboratory, value should be measured in research access, development velocity, experimental repeatability, data collection and time saved compared with building a humanoid platform from scratch.

NEO can be evaluated as an early consumer product and service. H2 EDU should be evaluated as research infrastructure. A direct ROI calculation using only purchase price would be meaningless.

Questions to Ask Before Buying 1X NEO or Unitree H2

  1. What exact tasks must the robot complete? List objects, locations, surfaces, payloads and acceptable completion quality.
  2. Does the task require a home product or a development platform? This single decision often resolves the comparison.
  3. Which configuration is included? Confirm hands, compute, controller, battery, charger, warranty and development access in writing.
  4. Is the task autonomous, teleoperated or hybrid? Ask for intervention rates and operator requirements.
  5. What happens when the robot fails? Define safe stop, dropped-object recovery, fall recovery and remote support.
  6. How is human safety validated? Request operating limits, fall zones, speed restrictions and relevant test documentation.
  7. How is household or research data handled? Clarify video, audio, telemetry, retention, model training and remote access.
  8. What is the delivered price? Include tax, freight, customs, accessories, compute, hands and service.
  9. When will the exact configuration ship? Do not use a general production announcement as an individual delivery guarantee.
  10. What is the service route? Identify local support, spare parts, repair location and turnaround.
  11. What software access is included? Confirm SDK, APIs, ROS compatibility, simulation and low-level control.
  12. What capabilities are current? Separate shipping functions from demonstrations and roadmap items.
  13. Can the robot complete a representative acceptance test? Use the real environment and objects before final commitment.
  14. Would a simpler robot solve the task better? Compare a mobile manipulator, cobot, service robot or fixed system.

Use the Anton Robots finder to identify the right robot category and models for your environment, budget and required level of autonomy.

Who Should Consider 1X NEO?

NEO belongs on the shortlist when most of the following are true:

  • You want a robot designed specifically for a home
  • You are comfortable being an early adopter
  • You understand that autonomy will improve over time
  • You accept scheduled expert assistance for unsupported tasks
  • Soft physical design and lower mass are high priorities
  • You need advanced manipulation of varied household objects
  • You value voice, app and remote-control interfaces
  • You want automatic charging
  • You are located in or can access the initial US delivery and support programme
  • You are willing to review detailed privacy and remote-access terms
  • Your expectation is assistance rather than immediate replacement of all household labour

Review the complete 1X NEO price, specifications and availability profile and the detailed 1X NEO review. The latest NEO 25-DoF hands update explains the most important recent hardware change.

Who Should Consider Unitree H2?

H2 belongs on the shortlist when most of the following are true:

  • You represent a university, laboratory, robotics company or technical organisation
  • You need a full-size humanoid platform
  • Your work focuses on motion, locomotion, control or embodied AI
  • You can operate the robot in a controlled space with trained personnel
  • You can maintain safe separation from people
  • You are prepared to purchase H2 EDU for secondary development
  • You can specify and integrate the required hands and compute
  • You can manage import, servicing and spare parts
  • You have engineering resources for integration and troubleshooting
  • You do not expect a turnkey autonomous home assistant

Review the complete Unitree H2 price, specifications and availability profile and the Unitree manufacturer profile before requesting a configuration-specific quote.

Common 1X NEO vs Unitree H2 Comparison Mistakes

  • Assuming both robots target the same buyer: NEO is a home product; H2 is primarily a research and demonstration platform.
  • Comparing $20,000 with $29,900 as if the packages match: standard H2 excludes development access and dexterous hands.
  • Assigning H2 EDU features to standard H2: secondary development is not included on the base model.
  • Assigning H2 Plus features to H2: H2 Plus is a separate, more expensive product.
  • Ignoring the NEO hand update: 1X announced 25-DoF shipping hands on 9 July 2026, although an older product table still lists 22.
  • Adding every DoF number into a single misleading total: body and hand joints are reported differently by the manufacturers.
  • Calling NEO fully autonomous: early robots arrive with basic autonomy and scheduled expert support.
  • Calling H2 an autonomous household robot: Unitree does not publish a supported home-task package.
  • Using peak payload as continuous payload: H2’s approximately 15 kg figure is a peak arm rating, while NEO separates arm payload, carry and lift.
  • Assuming 2070 TOPS equals 2070 FP4 TFLOPS: compute claims use different conventions and H2 configuration must be confirmed.
  • Transferring H1 or G1 speed to H2: Unitree does not publish a current H2 maximum speed on the product table.
  • Ignoring mass and body construction: a 70 kg rigid humanoid and a 30 kg soft humanoid create very different safety cases.
  • Assuming IP68 hands make the whole NEO waterproof: the body is rated IP44.
  • Judging autonomy from a polished demonstration: determine whether the action was autonomous, scripted, motion-captured or teleoperated.
  • Ignoring privacy: cameras, microphones, memory and remote access inside a home require explicit data controls.
  • Ignoring landed cost: H2 freight, customs, hands, compute and service may exceed the base price materially.
  • Buying a humanoid before comparing simpler robots: many tasks can be solved more safely and cheaply by a specialised platform.

FAQs

Is 1X NEO better than Unitree H2?

NEO is better for home use, consumer interaction, integrated dexterity, passive physical safety, self-charging and productised autonomy. It is not better for a laboratory that needs secondary development, custom compute and low-level experimentation.

Is Unitree H2 better than 1X NEO?

H2 EDU is better for full-size humanoid research, motion control and custom embodied-AI development. Standard H2 is not automatically better because it is larger or more powerful; it lacks secondary development and dexterous hands as standard.

Which robot should I buy for my home?

1X NEO. It is designed for domestic environments and includes a soft body, lower mass, tactile hands, voice interaction, autonomous charging and a consumer service model. H2 should not be treated as an unsupervised home appliance.

Which robot should a university buy?

Unitree H2 EDU is the stronger choice when the university needs custom development, locomotion research, motion control or embodied-AI experimentation. The standard H2 should not be purchased for that purpose without confirming its access limitations.

How much does 1X NEO cost?

1X lists Early Access ownership at $20,000 and a standard subscription at $499 per month. A $200 refundable deposit is due when ordering. Taxes, delivery, region and service terms may affect final cost.

How much does Unitree H2 cost?

The standard Unitree H2 is listed at $29,900 before tax and shipping. H2 EDU is priced by quotation, and optional hands, compute, customs and delivery can raise the total significantly.

Can you buy 1X NEO now?

You can place a pre-order through 1X’s order programme. First US deliveries are scheduled to begin in 2026, but 1X states that some customers will receive robots later as production and validation scale.

Can you buy Unitree H2 now?

Unitree lists H2 for sale and directs buyers to its sales channels. Actual stock, configuration and delivery time should be confirmed for the destination country.

Which robot is cheaper?

NEO has the lower published ownership price at $20,000 versus $29,900 for standard H2. The packages are not equivalent because H2 development access and dexterous hands require a different configuration.

Which robot is stronger?

H2 is the more powerful mechanical platform based on its published 120 N·m arm-joint and 360 N·m leg-joint maxima and approximately 15 kg peak arm payload. NEO publishes an 8.2 kg arm payload, 25 kg carry capacity and 70 kg lift claim, but these are different operating conditions.

Which robot has better hands?

NEO as supplied. Its latest announced hands have 25 force-controlled DoF each, tactile and shear sensing, IP68 sealing and fine manipulation capability. Standard H2 has no dexterous hands; H2 EDU supports optional models.

Does NEO have 22 or 25 DoF hands?

1X’s current order table still lists 22 DoF per hand, but its 9 July 2026 engineering announcement states that every NEO will ship with a new 25-DoF hand: 22 fully actuated finger and palm joints plus three wrist joints. Buyers should confirm the final order configuration.

How many degrees of freedom does Unitree H2 have?

Unitree publishes 31 body joints: six per leg, seven per arm, three at the waist and two in the head. Optional dexterous-hand DoF depend on the hand selected for H2 EDU.

Which robot is lighter?

NEO. It weighs approximately 30 kg, compared with approximately 70 kg for H2.

Which robot has longer battery life?

NEO publishes up to four hours, while H2 publishes approximately three hours. Actual runtime depends on activity, payload, compute and operating conditions.

Which robot has the larger battery?

H2 has a 972 Wh battery, compared with 842 Wh for NEO. Its published runtime is shorter because battery capacity alone does not determine efficiency.

Can NEO charge itself?

Yes. 1X states that NEO manages its battery and plugs itself in when charging is required.

Can Unitree H2 charge itself?

Unitree lists a quick-release smart battery and supplied charger but does not publish an automatic docking or self-charging function for H2.

Is 1X NEO fully autonomous?

Not at launch in the unrestricted sense. 1X says early NEOs arrive with foundational autonomy and improve over time. Complex unsupported tasks can use scheduled 1X Expert supervision.

Is Unitree H2 autonomous?

H2 has onboard computing, sensing and OTA motion capabilities, but Unitree does not publish a general autonomous task catalogue equivalent to NEO’s consumer service. Autonomy depends on software, configuration and development work.

Does NEO use teleoperation?

Yes. Owners can remotely control NEO, and scheduled 1X Experts can supervise complex tasks that the robot has not yet learned.

Does H2 support secondary development?

H2 EDU does. Unitree’s specification table marks secondary development as unavailable on the standard H2.

Does H2 come with dexterous hands?

The standard H2 does not. H2 EDU supports multiple optional dexterous-hand models, which must be specified and priced separately.

Does NEO support ROS 2?

1X does not position the consumer NEO purchase as an open ROS 2 or low-level joint-control development kit. Buyers requiring those interfaces should request written confirmation rather than assume access.

Which robot is safer around people?

NEO has the stronger safety-oriented physical design: lower mass, soft covering, low-inertia tendon drives, backdrivability and covered joints. H2’s manufacturer tells users to maintain sufficient distance because of its powerful structure.

Can H2 be used at home?

A technically skilled buyer could physically operate H2 indoors, but it is not presented as a safe, autonomous consumer home robot. Its mass, rigid body, safety warning and development orientation make it unsuitable as a direct NEO replacement.

Can NEO be used for research?

NEO may be interesting to researchers studying domestic robotics and human-robot interaction, but the consumer product is not positioned as an open development platform. Access to low-level software and data must be confirmed with 1X.

Which robot has more AI compute?

A responsible winner cannot be declared from the headline numbers. NEO lists up to 2070 FP4 TFLOPS on Jetson Thor. H2 marketing references 2070 TOPS, while its table lists standard and optional compute configurations. Units, precision and included hardware differ.

Which robot is faster?

NEO publishes 1.4 m/s walking and up to 6.2 m/s running. Unitree does not publish a directly comparable maximum H2 speed on the current product table, so a verified speed winner cannot be declared beyond available data.

Which robot is better for manipulation?

NEO is better as a complete purchased system because its advanced tactile hands are integrated with the robot. H2 EDU can become a powerful manipulation research platform, but the result depends on optional hands, compute and software.

Which robot is better for embodied AI?

NEO is better for experiencing a manufacturer-managed generalist AI product in homes. H2 EDU or H2 Plus is better for teams that want to develop, train and integrate their own embodied models.

Does NEO work without the internet?

1X states that Redwood runs onboard, but remote expert assistance, monitoring, updates and some connected features require connectivity. Buyers should confirm which core functions remain available offline.

What warranty does NEO include?

The Early Access ownership offer lists a three-year warranty and premium support.

What warranty does Unitree H2 include?

Unitree lists eight months for standard H2 and 12 months for H2 EDU, subject to the detailed product warranty terms.

Should I buy NEO or wait for a later version?

Buy NEO only if you value early access and accept evolving autonomy, delivery uncertainty and a developing support ecosystem. Waiting may provide more mature software, more field evidence and broader regional availability.

Should I buy standard H2 or H2 EDU?

Choose standard H2 only when supported motions, demonstrations and platform access meet the requirement. Choose H2 EDU when you need secondary development, optional dexterous hands, custom compute or serious research use.

Final Verdict

1X NEO wins the 1X NEO vs Unitree H2 comparison for consumers and home use. Unitree H2 EDU wins for robotics research and custom development.

NEO is the more coherent complete product. Its 30 kg soft body, tendon drives, covered joints, tactile hands, domestic AI, voice interface, remote support, four-hour runtime and self-charging all serve the same mission: placing a useful humanoid around ordinary people in ordinary homes. It is also less expensive on the published ownership price and offers a consumer subscription path.

Its limitations are equally important. NEO is an early product, delivery timing varies, independent field data remain limited and initial autonomy is foundational rather than universal. Scheduled expert assistance can bridge capability gaps, but it introduces privacy, connectivity and service questions that every buyer should understand.

H2 is a more powerful mechanical machine. Its 31 body joints, 70 kg full-size structure, 120 N·m arm-joint torque, 360 N·m leg-joint torque and configurable EDU route make it a compelling research platform at a comparatively accessible starting price. But the $29,900 standard model does not include the two features many technical buyers will assume it has: secondary development and dexterous hands. Those require H2 EDU or another configuration.

The correct 2026 framing is therefore:

  • NEO is a home robot product entering early consumer delivery.
  • H2 is a powerful humanoid platform whose real research value sits primarily in the EDU configuration.

Do not choose H2 because it is taller, heavier or has higher peak torque if the robot must operate safely in a home. Do not choose NEO because it has advanced hands and onboard AI if the project requires unrestricted low-level development.

Review 1X NEO and Unitree H2 in more detail, compare both robots side by side, explore the best humanoid robots in 2026, or use the Anton Robots finder to identify the right robot for your application and budget.

Primary Sources

Related Insights

Comparisons

UR5e vs Dobot CR5A: Which 5 kg Cobot Is Better?

Comparisons

Universal Robots UR10e vs FANUC CRX-10iA/: Full Cobot Comparison

Comparisons

Unitree Go2 vs Xiaomi CyberDog 2: Which Is Better?

Comparisons

Unitree G1 vs Unitree H2: Price, Specs & Verdict

Looking for a Robot for Your Business?

Find the right robot based on your application, industry and requirements, or explore and compare available models.

Browse Robots

GET STARTED

Find My Robot

Compare Robots

BROWSE BY TYPE

Humanoid Robots

Robot Dogs

Robotic Arms

Cobots

AMR Robots

AGV Robots

Service Robots

Companion Robots

BROWSE BY APPLICATION

Material Handling

Palletizing

Pick and Place

Welding

Inspection

Cleaning

Delivery

Security

BROWSE BY INDUSTRY

Manufacturing

Warehousing

Medical

Restaurants

Agriculture

Construction

Retail

Education