R Robot Review Desk

Buyer's guide · humanoid robots

Humanoid Robots in 2026: What You Can Actually Buy, Preorder, or Only Watch

The honest state of humanoid robotics in 2026: what's shipping, what's demo-ware, real prices, and who's furthest along.

By Max Langley ·

Disclosure: We earn commissions from links on this page, as an Amazon Associate we earn from qualifying purchases, at no extra cost to you. This never affects what we recommend. Read our editorial standards →

Most accessible humanoid you can buy today

Unitree Robotics

Unitree G1

Model: G1

The Unitree G1 is the most affordable production humanoid you can order right now without a corporate sales conversation (Unitree's larger H1 and smaller R1 are also self-serve orderable; the G1 is the value pick). Unitree shipped 5,500+ units in 2025 and is targeting 20,000 in 2026. The G1 is a research and early-industrial platform, not a household appliance, but it is genuinely purchasable, ships within weeks via RobotShop and the Unitree official store, and even appears on Amazon. The tradeoff: ~2-hour battery life, no integrated AI model for general tasks, and a product built for labs and developers, not consumers.

Check price at Unitree · From ~$16,000 (base); EDU configs from ~$43,900
Furthest along commercially (factory deployments)

Figure AI

Figure 03

Model: Figure 03

Figure AI has cleared the hardest gap in the category: from prototype to production at scale. Their BotQ facility went from 1 robot/day to 1 robot/hour in under 120 days, with 350+ Figure 03 units delivered as of April 2026. The BMW Spartanburg deployment (Figure 02) contributed to 30,000+ vehicles. Figure 03 is now expanding to additional BMW plants and is structured as a Robot-as-a-Service commercial contract at roughly $25/hour. Not a unit you can buy, but the closest thing to a commercially proven humanoid in the West. Figure is also targeting residential use cases and has built field-service infrastructure for home deployments.

Check price at Figure AI · ~$25/robot-operating-hour (RaaS model); not available for direct purchase
Consumer preorder open now

1X Technologies

1X NEO

Model: NEO

Norwegian startup 1X is the only company with open consumer preorders at a published price with a US delivery commitment in 2026. The NEO is purpose-built for the home, lighter and quieter than industrial humanoids, with privacy features including configurable no-go zones and face blurring. Day-1 tasks: door operation, fetching items, basic household assistance. 1X has raised significant backing including from OpenAI. The honest caveat: this is an early-access product from a startup, not a mature appliance. Buyers are funding the product's development as much as receiving a finished product.

Check price at 1X Technologies · $20,000 purchase (Early Access); or $499/month subscription. $200 deposit to reserve.
Most commercially proven in automotive

Apptronik

Apollo

Model: Apollo

Apptronik Apollo is deployed at Mercedes-Benz factories and GXO Logistics warehouses, backed by Google and Mercedes-Benz, and has raised $935M+ in its Series A. Apollo CEO Jeff Cardenas has stated a goal of under $50,000 per unit, but that's a target, not a current price, and units are not available outside enterprise partnerships. The $520M raise in February 2026 (at a $5.5B valuation) suggests serious commercial momentum. Most credible for large-company logistics and manufacturing procurement, not individual purchase.

Check price at Apptronik · Target <$50,000 per unit; RaaS pricing also offered. Not available for general purchase.
Internal only, not for sale (yet)

Tesla

Optimus Gen 3

Model: Optimus Gen 3

Tesla has over 1,000 Optimus Gen 3 robots working in its own factories as of early 2026: battery sorting, parts handling, quality inspection. This is the largest internal humanoid fleet deployed by any company. But it is emphatically not available externally. Tesla has not opened preorders, has not confirmed a B2B sales program, and analyst estimates for external availability range from late 2027 (limited B2B) to 2028+ (broad commercial). The $20K consumer price target from Musk is aspirational; most analysts put realistic consumer pricing at $30K+ when it eventually ships. The manufacturing story is real; the purchase story is not yet.

Check price at Tesla · Not for sale. Long-term consumer target: under $20,000 (Musk; disputed by analysts)
Shipping to enterprise partners only

Boston Dynamics

Atlas (Electric)

Model: Atlas (Electric)

Boston Dynamics' electric Atlas, distinct from the retired hydraulic Atlas, began shipping commercial units in 2026. All 2026 production is committed to Hyundai's Robotics Metaplant Application Center (RMAC) and Google DeepMind; additional customers are not expected until 2027. Atlas carries the most sophisticated locomotion and manipulation capabilities of any shipping humanoid, but it is inaccessible to anyone outside Boston Dynamics' existing enterprise pipeline, and pricing estimates vary wildly. The most interesting robot in the category for capability; the least accessible for procurement.

Check price at Boston Dynamics · Not publicly priced. Estimated $150,000–$420,000 per unit (industry analysts). 2026 units committed to Hyundai and Google DeepMind.

The honest framing for 2026

Humanoid robots are no longer exclusively lab demos. Some are on factory floors doing real work with documented KPIs. One is available for consumer preorder with a published price and a delivery timeline. But the gap between “this is real” and “you can buy one and use it at home” is still large, and the marketing around this category consistently obscures that gap.

This guide tries to close it. For each major platform we answer the same three questions: Is it shipping? To whom? At what price and under what terms?

What’s actually shipping to real customers

Figure 03 is the clearest commercial story in 2026. Figure AI’s BotQ manufacturing facility went from producing one robot per day to one per hour in under 120 days, and has delivered over 350 Figure 03 units as of late April 2026 (Figure AI, April 2026). The prior generation, Figure 02, completed an 11-month BMW Spartanburg deployment that contributed to 30,000+ X3 vehicles, running 1,250+ hours with 90,000+ parts loaded (Figure AI, November 2025). These are not demo numbers. They are production KPIs with documented cycle times and placement accuracy requirements. The commercial model is Robot-as-a-Service at roughly $25/robot-operating-hour, not a unit you can purchase outright. Figure is not available for individual or small-business procurement.

Apptronik Apollo is deployed at Mercedes-Benz assembly plants and GXO Logistics warehouses, with Google and Mercedes as lead backers. A $520M Series A raise in February 2026, bringing total financing above $935M, reflects genuine enterprise traction. Apollo’s CEO has cited a target unit price under $50,000, but that is a goal, not a current price, and procurement requires an enterprise partnership discussion, not a product page.

Boston Dynamics Atlas (electric) began shipping commercial units in 2026. All current production is committed to Hyundai’s Robotics Metaplant Application Center and Google DeepMind. Additional customers are not expected to be onboarded until 2027. Pricing is not publicly disclosed; analyst estimates range widely, from roughly $150,000 to over $400,000 per unit.

Unitree G1 is the outlier: a full-size bipedal humanoid you can order online today, without a sales call, starting at approximately $16,000. Unitree shipped over 5,500 units in 2025 and targets 20,000 in 2026, and the G1 is available through the Unitree official store, RobotShop, and Amazon. EDU configurations with expanded degrees of freedom run from $43,900 to $73,900. The G1 is a research and developer platform, not a household robot, but it is genuinely, immediately purchasable.

The consumer preorder story: 1X NEO

1X NEO is the only humanoid with open consumer preorders at a fixed price and a US delivery promise in 2026. The Norwegian company, backed partly by OpenAI, is asking $200 to reserve a place in line toward a $20,000 purchase price (or a $499/month subscription). First deliveries are targeted for US early-access customers in 2026, with broader markets in 2027 (1X order page).

The NEO is purpose-designed for the home, lighter and quieter than industrial platforms, with privacy features including configurable no-go zones and the ability to blur faces of people it interacts with. Stated Day 1 capabilities are conservative and honest: opening doors, fetching items, turning off lights. The company is not claiming general domestic labor.

The important caveat: 1X is a startup shipping an early-access product. Buyers at this price and this stage of development are effectively participating in the product’s development cycle. That is not a dealbreaker if you understand what you’re getting, but it is meaningfully different from buying a consumer appliance.

The Tesla Optimus situation

Tesla has more humanoid robots working in production environments than any other company: over 1,000 Optimus Gen 3 units deployed across its own factories as of early 2026, performing battery sorting, parts handling, and quality inspection. The Fremont facility has been partially converted for Optimus Gen 3 manufacturing.

What Tesla does not have: an external sales program, open preorders, or a confirmed B2B commercial offering for non-Tesla customers. As of mid-2026, Optimus is internal only. Elon Musk’s stated long-term consumer price target of under $20,000 is aspirational; analysts generally place realistic external availability at late 2027 for limited B2B pilots, with broader consumer access in 2028 at the earliest. Third-party sites offering to “pre-sell” Tesla Optimus are not affiliated with Tesla.

The manufacturing achievement is real and significant. The availability story is not.

What separates the commercial leaders from the demos

A useful framework: the difference between a robot that performs in controlled demos and one that performs commercially is the KPI conversation. Figure’s BMW deployment had three measurable targets: cycle time (84 seconds total, 37 seconds load time), placement accuracy (>99% per shift), and interventions per shift (goal: zero). Those are the numbers that make a humanoid deployment a finance-defensible operating line rather than an R&D experiment.

Most humanoid companies in 2026 are still on the demo side of that line. The ones with real commercial deployments, Figure, Apptronik, Agility (at Amazon and GXO), and Boston Dynamics (at Hyundai), are the ones who have sat through those KPI negotiations with industrial customers and delivered.

The consumer side of the market is even earlier. The 1X NEO’s Day 1 task list (door operation, fetching, lights) is conservative not because of limited technology but because 1X has correctly identified that a household use case requires reliability across a chaotic, unpredictable environment that even the best factory robots struggle with. Getting a robot to load a car part in a structured fixture at 99% accuracy is genuinely easier than getting one to reliably put a glass in the dishwasher across 1,000 different kitchens.

Who should be watching this space vs. acting on it

Watch, don’t buy yet: Anyone looking for a household assistant that works like a capable human. That product does not exist in 2026. The 1X NEO is the most credible attempt, and even 1X frames first-year use as a collaboration between user and product.

Consider now, with clear eyes: Researchers, developers, and robotics labs: the Unitree G1 is a legitimate and accessible platform at its price point. Large enterprises in automotive, logistics, and manufacturing: Figure, Apollo, and Boston Dynamics Atlas are operational and investable, if not purchasable off-the-shelf.

The price to watch: When a full-size bipedal humanoid with a general AI model crosses below $10,000, the consumer calculus changes. That is not a 2026 event. The G1 at $16,000 is the current floor, and it is a developer tool.

The vaporware filter

Not every humanoid company with a slick video is a company with a shipping product. The signal to look for: documented commercial deployment with named customers and production KPIs. Figure, Apptronik, Agility, Boston Dynamics, and Unitree have that. Many others do not. A convincing demo reel and a waiting list are table stakes in this category, not evidence of commercial readiness.

Frequently asked questions

Can a regular consumer actually buy a humanoid robot in 2026?
Technically yes. The Unitree G1 can be ordered online from around $16,000 and ships in weeks. The 1X NEO takes a $200 deposit against a $20,000 purchase price with 2026 US delivery promised. But neither is a household appliance in the sense most buyers expect. The G1 is a developer/research platform. The NEO is an early-access product from a startup where you are participating in development as much as receiving a finished product. If your frame of reference is a Roomba or a smartphone, nothing is in that tier in 2026. If your frame of reference is a car (large purchase, some unreliability, requires skill to operate safely), the G1 is legitimately there.
What can today's humanoid robots actually do reliably?
The honest answer from real deployments is: narrow, structured tasks, performed reliably with significant setup. Figure 02 at BMW loaded sheet-metal parts within a 5mm tolerance on a structured assembly line, 90,000+ parts over 1,250+ hours. That task was defined, bounded, and practiced extensively. The same robot could not walk into a new factory and start working autonomously. As of 2026, humanoids can perform pick-and-place manufacturing tasks, navigate factory floors and handle light logistics, and execute household tasks like opening doors or fetching items (1X NEO's stated Day 1 use cases). What they cannot do reliably: adapt to novel environments without significant configuration, perform dexterous manipulation of unfamiliar objects, or operate for more than a few hours without supervision or recharging.
Is Tesla Optimus the most advanced humanoid?
Tesla has the largest internal deployment (1,000+ units in its own factories) and the most aggressive production targets. But 'most advanced' depends on what you measure. Boston Dynamics Atlas has the most impressive locomotion and manipulation demos. Figure 03 has the most documented real-world commercial deployment with rigorous KPIs (placement accuracy, cycle time, interventions per shift). 1X has the most consumer-oriented design. Unitree has the lowest price for a functional bipedal platform. Tesla's advantage is manufacturing scale and vertical integration. They build their own actuators and AI chips, but Optimus has never been evaluated in a third-party deployment, so independent comparison is impossible.
Why are humanoid robots so expensive?
Actuators. A human-scale bipedal robot requires 20–40+ degrees of freedom, each requiring a motor, gearbox, encoder, and controller. Unlike industrial robot arms (which have 6 axes and have been manufactured at scale for 50 years), humanoid actuators are custom, low-volume, and expensive to qualify for reliability. Unitree has driven the G1 to $16,000 partly by manufacturing actuators in massive volume across its dog-robot line as well. The long-term $20,000 consumer price point that Musk and others cite assumes actuator costs falling 10–20x as volume scales, which is plausible on a 5–10 year horizon but is not where the market is in 2026.
What about companies like Agility (Digit) and UBTECH?
Agility Robotics' Digit is in active commercial deployment at Amazon and GXO Logistics warehouses and is one of the most widely deployed warehouse humanoids in 2026. UBTECH's Walker S2 is similarly deployed at automotive plants in China (BYD, Nio). Neither company is targeting consumer sales or individual purchase; both are enterprise logistics and manufacturing platforms. They represent the same commercial tier as Figure and Apollo: proven in structured environments, unavailable outside enterprise partnerships, and not independently priced for the public.
When will a humanoid robot be available at a 'normal' consumer price?
The most optimistic credible timeline: $30,000–50,000 enterprise units broadly available 2027–2028, with mass-market consumer pricing ($10,000–20,000) not before 2030 at the earliest. Goldman Sachs has revised its humanoid market forecast upward multiple times, but the 2026 baseline is still 'early commercial industrial,' not 'consumer electronics.' The 1X NEO at $20,000 is the closest thing to a consumer product, and its price reflects early-production costs and startup risk premium, not the long-run price floor.

Sources

Every claim in this guide that isn't first-person experience is traceable to one of the sources below. URLs verified at publication; some may rot. Let us know if so.

  1. Ramping Figure 03 Production · Figure AI, April 29, 2026Primary source. BotQ facility producing 1 Figure 03/hour; 350+ units delivered; 24x throughput improvement in under 120 days.
  2. F.02 Contributed to the Production of 30,000 Cars at BMW · Figure AI, November 19, 2025Primary source. Figure 02 BMW Spartanburg deployment results: 90,000+ parts loaded, 1,250+ hours runtime, contributed to 30,000+ X3 vehicles. KPIs documented: 84-second cycle time, >99% placement accuracy target.
  3. Unitree G1: Official Product Page · Unitree Robotics, 2026Primary source. G1 available for direct purchase, prices listed in USD. Ships within weeks.
  4. Unitree G1 Humanoid Drops for $16,000, Upending the Robotics Market · RoboHorizon, April 2026Market context on Unitree G1 pricing and 2026 shipment targets (20,000 units). IPO context ($610M Shanghai filing).
  5. 1X NEO Home Robot: Order Page · 1X Technologies, 2026Primary source. $200 deposit, $20,000 purchase or $499/month subscription. 2026 US delivery commitment.
  6. NEO humanoid designed for household use, available for preorder · The Robot Report, 2026Independent coverage of 1X NEO preorder launch, pricing, and US delivery timeline.
  7. Apptronik raises $520M to produce humanoid Apollo robot commercial deployments · SiliconANGLE, February 11, 2026$520M raise at $5.5B valuation; total Series A exceeds $935M. Google and Mercedes-Benz as lead backers. Apollo deployed at Mercedes-Benz factories and GXO Logistics.
  8. Apollo: Apptronik Product Page · Apptronik, 2026Primary source. Product overview, deployment capabilities, partnership context.
  9. CES 2026: Boston Dynamics Set to Ship First Atlas Humanoids This Year · Automate.org / A3, January 2026Boston Dynamics electric Atlas beginning production in 2026; first units to Hyundai RMAC and Google DeepMind. Additional customers not until 2027.
  10. Tesla Optimus Price & Availability 2026: What We Know · Silicon Valley Robotics Center, 2026Synthesis of Tesla Optimus deployment status, pricing claims, and availability timeline. Notes 1,000+ internal units deployed; no external sales as of mid-2026.
  11. Humanoid Robotics 2026: Figure, Optimus, 1X Commercial Reality · VaaSBlock, 2026Market survey of commercial status across major humanoid platforms; pricing context for enterprise deployments.
  12. Figure AI Stock: $39B Valuation. Is It a Buy? · TSG Invest, 2026Financial context on Figure AI valuation, IPO outlook, and commercial deployment revenue structure.