Updated September 2026
Welcome to the most comprehensive archive of humanoid robots in development and deployment worldwide. From Tesla's Optimus to China's high-volume robotics ecosystem, this living database tracks the machines reshaping manufacturing, logistics, healthcare, and domestic life. The humanoid build-out is no longer a preview: thousands of units are already in the field, commercial pilots have become multi-site contracts, and production lines—not just prototypes—are now the story.
The race to deploy humanoids at scale is dominated by companies with paid pilots, multi-year logistics contracts, and factory partnerships. These are no longer only research platforms—they are robots moving totes, sequencing parts, and collecting the data that trains the next generation.
Tesla's Optimus program remains the industry's most ambitious volume bet, but the 2025 consumer timeline did not hold. Generation 3 is in low-volume internal production in 2026, with Fremont lines being converted and a dedicated Texas factory still under construction for a high-volume target around summer 2027. The headline hardware upgrade is the hand: 22 degrees of freedom per hand, tendon-driven, with actuators moved into the forearm. Circulating body specs remain about 5'8" (173 cm) and 125 pounds (57 kg). Tesla still cites a long-term consumer price of $20,000–$30,000; there is no public waitlist, and outside availability is generally described as late 2026 enterprise trials and 2027 consumer timing. Earlier talk of 5,000–10,000 units by the end of 2025 was missed. Optimus is the benchmark for scale ambition—not yet for shipped customer fleets.
After retiring hydraulic Atlas, Boston Dynamics moved from the 2024 all-electric prototype to a production machine unveiled at CES in January 2026. That robot stands about 1.9 meters (6.2 feet) tall and 90 kg (198 pounds), with 56 degrees of freedom, fully rotational joints, a 2.3-meter reach, and a rated lift of 50 kg instant / 30 kg sustained. Runtime is about four hours with autonomous battery swap. Hyundai is the industrial anchor: 2026 production is committed to Hyundai's Robotics Metaplant Application Center and Google DeepMind, with broader line-side work at the Georgia Metaplant planned later in the decade. Early use is parts sequencing, not a full welding-and-inspection rollout. Atlas is now a commercial product in limited allocation—not a research demo with unsettled specs.
Unitree still sets the price floor. The G1 base model lists at $13,500, with street and dealer prices often $16,000–$18,000 and EDU configurations much higher. Official standing height is 132 cm; weight is about 35 kg; degrees of freedom run 23 on the base unit and up to 43 on EDU builds; walking speed is about 2 m/s. In September 2026 Unitree introduced G1+, adding neck degrees of freedom, stronger shoulder/waist/arm motors, binocular-plus-wide-angle vision, and an external power port. China list price for the standard G1+ is about ¥95,000. The platform remains the default lab and education humanoid, not a factory workhorse.
Agility still holds the clearest Western logistics proof point. Digit began as the industry's first multi-year humanoid Robotics-as-a-Service contract at a GXO site serving Spanx, and has since accumulated tens of thousands of operating hours across customers including Amazon, Mercado Libre, Schaeffler, and Toyota. Digit 4 is the bird-legged generation that proved tote handling; Digit 5, with more human-like legs and longer runtime, is the 2026 product Agility is taking public-market and factory-scale. The robot remains a warehouse specialist—moving totes between AMRs and conveyors across eight-hour-plus shifts—not a general home assistant.
Humanoids are entering plants because they can use aisles, carts, and fixtures built for people. The 2026 pattern is narrower than the pitch decks: tote moves, sheet-metal loading, and parts sequencing first; unstructured assembly later.
Figure's BMW work is still the best-documented Western factory case. Figure 02 spent roughly ten to eleven months at Spartanburg, supporting production of more than 30,000 BMW X3 vehicles, loading sheet-metal parts with millimeter-class precision. Figure retired that generation and launched Figure 03 on 9 October 2025 around Helix, its vision-language-action stack. Hardware changes include a faster, wider camera suite, tactile fingertips, palm cameras, soft goods, and wireless charging. In June 2026 BMW put Figure 03 onto logistics sequencing in Hall 52. In September 2026 Figure reported Helix 2.5 completing chores such as towel folding, bed making, and tidying across 30 previously unseen Bay Area homes. Figure 03 is now both a plant robot and a home-research platform; it is not a consumer SKU with a public price.
Canadian firm Sanctuary AI built Phoenix as a cognitive humanoid with high-torque arms and highly dexterous hands, including needle-threading demos and teleoperation-led task learning. That hardware lineage continues (Phoenix Gen 8), but the commercial story changed in mid-2026. Sanctuary is now selling Physical AI and dexterous hands onto existing industrial robots, after posting production-rate results on a conveyor wire-plugging task at a Tier 1 automotive supplier. Treat Phoenix as a research and data platform, not as a competing shipping humanoid in the same class as Digit or Figure 03.
The home is still the next battleground. Only one company is taking refundable consumer deposits against 2026 U.S. deliveries; everyone else is in factory, alpha, or campus use.
Norwegian startup 1X opened NEO pre-orders in late October 2025—the first widely marketed consumer home humanoid with a live order page. Pricing is still $20,000 Early Access ownership or $499 per month, with a $200 refundable deposit and U.S. deliveries advertised to start in 2026. NEO is a light, compliant design (about 66 pounds, tendon drive, soft lattice body) meant to open doors, fetch items, and learn chores with a human in the loop. Complex tasks still lean on scheduled teleoperation while Redwood / World Model autonomy matures. As of late 2026, large-scale customer deliveries are not yet a settled public fact; CEO commentary now points at 2027 volume (a 50,000-unit shipment target) and California factory capacity well above 2026 output. NEO is the consumer beachhead—not a fully autonomous housekeeper.
EV maker XPeng used IRON's unusually smooth gait to argue that human-like walking is a product feature, not a party trick. That story has moved off the stage. In September 2026 XPeng commissioned an automated IRON line in Guangzhou and showed a unit walking off it after assembly. Current specs: 76 degrees of freedom, 21 per hand, three in-house Turing chips, and an enclosed flexible lattice shell. Mass production is targeted by the end of 2026, with first robots going into XPeng stores and campuses before a 2027 commercial launch. IRON is no longer “a Q1 2026 Physical AI teaser.” It is a vehicle-company humanoid entering manufacturing.
These platforms still push materials, control, and embodied AI. Some will become products; others stay testbeds.
NVIDIA's Isaac GR00T stack remains the default infrastructure layer many humanoid makers train and deploy against. Foundation-model work has split across vendors: Figure's Helix 2.5, Tesla's FSD-derived networks plus Grok, XPeng's on-robot Physical AI, and Google DeepMind's robotics models on partners including Atlas and Apollo. The 2026 lesson is unchanged and sharper: hardware without a trainable whole-body policy does not stay in a plant.
The industry is no longer waiting for the first pilot. Mercedes-Benz, Jabil, Amazon, GXO, BMW, Hyundai, and a long list of Chinese manufacturers are either running robots or building the plants that will. The remaining constraints are familiar: battery endurance still clusters around two to eight hours, dexterity is improving because of vision-language-action models rather than more motors alone, and cost curves are being written in Hangzhou and Shanghai as much as in California. American companies still lead published third-party factory case studies and foundation-model demos. Chinese firms lead installations and bill-of-materials. European buyers still set the safety and workplace-integration bar.
The question for late 2026 is no longer whether humanoids leave the lab. It is which platforms survive contact with an eight-hour shift, which AI stacks transfer from one home or aisle to the next without a new dataset, and whether consumer deliveries in 2026–2027 can match the pre-order marketing. The robots are real. The deployments are uneven. The production lines are starting to run.
For complete summary, editorial, specifications, and additional news on humanoid robots, explore the Humanoid.Press Database—the industry's most comprehensive technical resource.
China's humanoid robotics sector exploded in 2024-2025, with dozens of startups and established tech firms flooding the market with affordable platforms. The Chinese government designated humanoid robotics as a strategic priority, channeling billions into R&D and manufacturing infrastructure. Companies like AgiBot, Fourier Intelligence, UBTECH, and Unitree are leading a wave of cost-competitive robots targeting both industrial and consumer markets.
The Chinese advantage: vertical integration, massive manufacturing scale, and aggressive pricing. While Western companies emphasize AI sophistication and safety certification, Chinese manufacturers prioritize rapid iteration and market deployment. This dual strategy is reshaping global competition, forcing American and European firms to accelerate timelines or risk obsolescence.
Unitree Robotics disrupted the research market with the G1, a full-featured humanoid priced starting at just $13,500 (base model)—one-tenth the cost of comparable platforms. Standing 132cm and weighing 35kg, the G1 packs 23-43 degrees of freedom (depending on configuration), 3D LiDAR, Intel RealSense cameras, force-controlled dexterous hands, and impressive athletic capabilities including backflips and 2 m/s walking speed. Universities worldwide adopted G1 units throughout 2025, making it one of the year's best-selling research platforms and proving that advanced bipedal robotics is no longer exclusive to well-funded labs.
Shanghai-based AgiBot raised significant funding in 2024-2025 to scale production of its Iingi X2 and A2 platforms. The company targets both industrial automation and service robotics, with pilots underway in manufacturing, logistics, and retail sectors across China. AgiBot represents the new generation of Chinese robotics firms: well-capitalized, technically sophisticated, and moving at breakneck speed from prototype to production.
American humanoid companies lead in AI integration, leveraging advances in computer vision, natural language processing, and reinforcement learning. Firms like Figure AI, Apptronik, Agility Robotics, and Tesla prioritize end-to-end neural networks and foundation models trained on massive datasets. This AI-first approach aims to create robots that can generalize across tasks rather than requiring task-specific programming.
Figure AI completed an 11-month pilot deployment at BMW's Spartanburg plant, where Figure 02 robots worked on active production lines loading sheet metal parts with 5-millimeter precision. The company then launched Figure 03 featuring Helix—a proprietary vision-language-action AI that fuses perception, reasoning, and manipulation. Figure 03's redesigned sensory suite offers double the frame rate and 60% wider field of view compared to its predecessor, enabling complex household tasks like folding laundry and loading dishwashers. The BMW partnership validated that humanoids can operate safely and productively in real manufacturing environments, providing critical data for next-generation designs.
Tesla's Optimus program represents the most ambitious consumer robotics bet in history. With target pricing between $20,000-$30,000 and plans for thousands of factory deployments in 2025-2026, Elon Musk envisions humanoids as ubiquitous as smartphones. The latest generation features 22 degrees of freedom in each hand, end-to-end neural networks borrowed from Full Self-Driving, and the ability to learn tasks through human observation. Tesla deployed Optimus internally throughout 2024-2025, targeting 5,000-10,000 units by year-end 2025. While production has faced delays, Optimus remains the industry's benchmark for scalable, affordable humanoids.
European humanoid development emphasizes workplace safety, regulatory compliance, and natural human-robot collaboration. Companies like Engineered Arts (UK), PAL Robotics (Spain), and projects across European research institutions design robots for stringent safety frameworks. While production volumes lag Asian and American competitors, European platforms lead in certification-ready designs for healthcare, hospitality, and regulated environments.
Engineered Arts' Ameca became the world's most recognizable humanoid through viral demonstrations of its remarkably expressive face and lifelike reactions. While not designed for industrial labor, Ameca excels at human-robot interaction research, entertainment, and public engagement. The platform's modular design and exceptional facial animation make it the gold standard for studying social robotics and human perception of artificial agents. Ameca has appeared at conferences, exhibitions, and media events worldwide, serving as an ambassador for the broader humanoid robotics industry.
Japan pioneered humanoid robotics with Honda's ASIMO and Sony's QRIO in the 1990s-2000s. Today, Japanese research institutions continue advancing bipedal locomotion, human-robot interaction, and android aesthetics through platforms like Alter, HRP-5P, and research androids from Hiroshi Ishiguro's laboratory. While commercial deployments lag China and the US, Japan's contributions to fundamental robotics research—from dynamic walking algorithms to uncanny valley studies—remain foundational to the global industry.
University labs worldwide develop specialized humanoids for research in locomotion, manipulation, embodied AI, and human-robot interaction. These platforms prioritize modularity, open-source compatibility, and experimental flexibility over commercial viability. Academic robots often pioneer techniques that eventually reach commercial products—from dynamic balance control to learning-based manipulation—making them crucial testbeds for the industry's future.
The race to deliver affordable, safe home humanoids intensified in 2025. Norwegian startup 1X Technologies opened pre-orders for NEO—the world's first consumer humanoid with confirmed 2026 delivery dates. Priced at $20,000 or $499/month subscription, NEO weighs just 66 pounds and uses soft, compliant actuation designed for safe household interaction. The platform will ship with basic autonomous capabilities but relies heavily on teleoperation and cloud-based learning. Other firms are developing home-oriented platforms emphasizing safety, quiet operation, and approachable aesthetics over industrial strength.
The humanoid robotics industry stands at a critical juncture. After decades of research and hype cycles, the technology is finally ready for commercial deployment. Battery efficiency, AI capabilities, manufacturing costs, and investor confidence have all reached thresholds enabling mass production. Major manufacturers from Mercedes-Benz to Amazon have confirmed large-scale pilots and deployments. The next 12-24 months will determine which companies capture market share, which platforms become industry standards, and whether consumers embrace humanoids in homes.
Key questions for 2026: Will Tesla achieve its ambitious Optimus production targets? Can 1X NEO deliver successfully to early adopters? Will Chinese manufacturers flood global markets with affordable alternatives? How quickly will manufacturing partnerships scale beyond pilots? And most importantly—will the public accept humanoid robots as coworkers and companions, or resist the transformation?
The robots are real. The deployments are happening. Major corporations are betting billions on embodied AI. The race to define the future of work, manufacturing, and home life is underway. After 30 years of science fiction promises, humanoid robots are finally transitioning from labs to production lines to living rooms. Welcome to the embodied AI era.