January 29, 2026
12 min read
A place for humanoid and AI robotics enthusiasts
March 5, 2025
14 min read
March 12, 2025
6 min read
Dean Fankhauser
January 29, 2026
12 min read
March 20, 2025
12 min read
January 29, 2026
12 min read
March 20, 2025
12 min read
January 23, 2025
3 min read
January 9, 2025
5 min read
Regular updates ensure that readers have access to fresh perspectives, making Poster a must-read.
March 5, 2025
14 min read
March 12, 2025
6 min read
January 9, 2025
5 min read
December 30, 2024
4 min read
February 2, 2026
18 min read



Join other humanoid and AI robotics enthusiasts in our invite-only community. Apply now to join.
Unitree G1 and H2 robots performed the world's first fully autonomous humanoid kung fu routine at China's 2026 Spring Festival Gala. With 20,000 units planned for 2026, here's what this means for the robotics industry.
On February 16, 2026, approximately 679 million people watched something unprecedented unfold on their screens: dozens of Unitree humanoid robots performing fully autonomous kung fu on the stage of China's Spring Festival Gala. No teleoperation. No pre-programmed dance moves. Just pure, AI-driven martial arts that included backflips, weapon handling, and a record-breaking 7.5-rotation Airflare spin.
This wasn't a tech demo in a sanitized laboratory. This was the Unitree Spring Festival 2026 moment—broadcast live to the largest television audience on Earth during China's equivalent of the Super Bowl.
The Spring Festival Gala (Chunwan) is China's most-watched annual broadcast, traditionally featuring music, dance, and cultural performances. In 2026, Unitree Robotics made history by debuting its Unitree G1 robot fleet alongside the larger H2 models in a segment titled "Cyber Real Kung Fu."
According to Unitree's official press release, this marked "the world's first fully autonomous humanoid robot cluster martial arts performance." The routine wasn't just impressive—it shattered multiple technical records:
The H2 models added dramatic flair, appearing in Monkey King armor and even riding Unitree's B2W quadruped robot dogs as "somersault clouds"—a reference to the legendary Chinese folk hero Sun Wukong.
The star of the show, the Unitree G1, represents Unitree's push into affordable humanoid robotics. Here are the key specifications that enabled those viral kung fu moves:
| Specification | Unitree G1 |
|---|---|
| Height | 127 cm (4.2 ft) |
| Weight | 35 kg (77 lbs) |
| Degrees of Freedom | 23+ (up to 43 with dexterous hands) |
| Max Walking Speed | 2+ m/s (over 7 km/h) |
| Battery Life | ~2 hours (quick-swap design) |
| Starting Price | $13,500 USD (base model) |
What sets the G1 apart isn't just hardware—it's the AI driving it. Unitree implemented systematic upgrades across algorithms, hardware, and systems specifically for the gala performance. The robots used reinforcement learning combined with force-position hybrid control, enabling the precise, fluid movements that captivated the global audience.
While the G1 handled the acrobatic kung fu sequences, Unitree's H2 model brought the theatrical presence. Standing taller and built for heavier industrial applications, the H2 appeared at both the Beijing main venue and the Yiwu sub-venue.
Priced at approximately $29,900, the H2 targets different use cases—warehouse logistics, manufacturing assistance, and heavy-duty manipulation tasks. Its appearance at the gala demonstrated that Unitree isn't just building research platforms; they're building a full product ecosystem for China humanoid robots 2026 and beyond.
Perhaps more significant than the viral performance was what Unitree founder Wang Xingxing announced afterward. In an interview with tech outlet 36Kr, Wang revealed that Unitree plans to ship between 10,000 and 20,000 humanoid robots in 2026.
To put this in perspective:
This isn't aspirational marketing—it's a signal that humanoid robots are transitioning from experimental technology to commercial products. When a company commits to shipping 20,000 units, supply chains, manufacturing processes, and quality control systems must already be in place.
If you're considering purchasing a humanoid robot—whether for research, education, or early commercial applications—the Unitree Spring Festival 2026 performance carries several implications:
Live performances don't lie. When robots execute complex martial arts routines autonomously in front of hundreds of millions of viewers, it validates the underlying technology in ways that controlled demos never can. The G1's performance proves it can handle dynamic, unpredictable scenarios—not just scripted laboratory tasks.
At $13,500 for the base G1, Unitree offers arguably the best value proposition in the humanoid market. Competitors like Boston Dynamics' Atlas remain research-only platforms without consumer pricing. Tesla's Optimus has yet to reach general availability. The G1 is shipping now.
Unitree's 20,000-unit production target means more robots in the field, more edge cases discovered, and faster iteration on reliability issues. Early adopters benefit from a company operating at scale rather than building one-off prototypes.
The gala showcased integration between Unitree's humanoid robots (G1, H2) and quadruped platforms (B2W). This ecosystem approach suggests long-term platform support, shared development tools, and interoperability—critical factors for anyone building robotics applications.
Unitree wasn't alone at the 2026 Spring Festival Gala. Other Chinese robotics companies including Galbot, Noetix, and MagicLab also featured robots in the broadcast, signaling a coordinated national effort to showcase domestic robotics capabilities.
China's government has identified humanoid robotics as a strategic technology priority, with provincial governments offering subsidies and incentives for robot manufacturers. The Spring Festival Gala appearance served dual purposes: entertaining domestic audiences while broadcasting China's robotics ambitions to the world.
For international buyers, this competitive landscape means more options, faster innovation, and—crucially—continued downward pressure on prices.
The humanoid robot kung fu performance at the 2026 Spring Festival Gala will be remembered as a watershed moment. Not because robots doing martial arts is inherently useful, but because it demonstrated capabilities that transfer directly to practical applications: dynamic balance, precise manipulation, real-time adaptation, and coordinated multi-robot operation.
Unitree has proven its robots can perform under pressure at the highest stakes imaginable. Now the question becomes: what will you build with one?
Whether you're a researcher, educator, or early commercial adopter, the Unitree G1 represents the most accessible entry point into humanoid robotics available today. Browse our complete selection of Unitree robots—including the G1, H2, and Go2 quadruped platforms—to find the right fit for your application.
On February 16, 2026, approximately 679 million people watched something unprecedented unfold on their screens: dozens of Unitree humanoid robots performing fully autonomous kung fu on the stage of China's Spring Festival Gala. No teleoperation. No pre-programmed dance moves. Just pure, AI-driven martial arts that included backflips, weapon handling, and a record-breaking 7.5-rotation Airflare spin.
This wasn't a tech demo in a sanitized laboratory. This was the Unitree Spring Festival 2026 moment—broadcast live to the largest television audience on Earth during China's equivalent of the Super Bowl.
The Spring Festival Gala (Chunwan) is China's most-watched annual broadcast, traditionally featuring music, dance, and cultural performances. In 2026, Unitree Robotics made history by debuting its Unitree G1 robot fleet alongside the larger H2 models in a segment titled "Cyber Real Kung Fu."
According to Unitree's official press release, this marked "the world's first fully autonomous humanoid robot cluster martial arts performance." The routine wasn't just impressive—it shattered multiple technical records:
The H2 models added dramatic flair, appearing in Monkey King armor and even riding Unitree's B2W quadruped robot dogs as "somersault clouds"—a reference to the legendary Chinese folk hero Sun Wukong.
The star of the show, the Unitree G1, represents Unitree's push into affordable humanoid robotics. Here are the key specifications that enabled those viral kung fu moves:
| Specification | Unitree G1 |
|---|---|
| Height | 127 cm (4.2 ft) |
| Weight | 35 kg (77 lbs) |
| Degrees of Freedom | 23+ (up to 43 with dexterous hands) |
| Max Walking Speed | 2+ m/s (over 7 km/h) |
| Battery Life | ~2 hours (quick-swap design) |
| Starting Price | $13,500 USD (base model) |
What sets the G1 apart isn't just hardware—it's the AI driving it. Unitree implemented systematic upgrades across algorithms, hardware, and systems specifically for the gala performance. The robots used reinforcement learning combined with force-position hybrid control, enabling the precise, fluid movements that captivated the global audience.
While the G1 handled the acrobatic kung fu sequences, Unitree's H2 model brought the theatrical presence. Standing taller and built for heavier industrial applications, the H2 appeared at both the Beijing main venue and the Yiwu sub-venue.
Priced at approximately $29,900, the H2 targets different use cases—warehouse logistics, manufacturing assistance, and heavy-duty manipulation tasks. Its appearance at the gala demonstrated that Unitree isn't just building research platforms; they're building a full product ecosystem for China humanoid robots 2026 and beyond.
Perhaps more significant than the viral performance was what Unitree founder Wang Xingxing announced afterward. In an interview with tech outlet 36Kr, Wang revealed that Unitree plans to ship between 10,000 and 20,000 humanoid robots in 2026.
To put this in perspective:
This isn't aspirational marketing—it's a signal that humanoid robots are transitioning from experimental technology to commercial products. When a company commits to shipping 20,000 units, supply chains, manufacturing processes, and quality control systems must already be in place.
If you're considering purchasing a humanoid robot—whether for research, education, or early commercial applications—the Unitree Spring Festival 2026 performance carries several implications:
Live performances don't lie. When robots execute complex martial arts routines autonomously in front of hundreds of millions of viewers, it validates the underlying technology in ways that controlled demos never can. The G1's performance proves it can handle dynamic, unpredictable scenarios—not just scripted laboratory tasks.
At $13,500 for the base G1, Unitree offers arguably the best value proposition in the humanoid market. Competitors like Boston Dynamics' Atlas remain research-only platforms without consumer pricing. Tesla's Optimus has yet to reach general availability. The G1 is shipping now.
Unitree's 20,000-unit production target means more robots in the field, more edge cases discovered, and faster iteration on reliability issues. Early adopters benefit from a company operating at scale rather than building one-off prototypes.
The gala showcased integration between Unitree's humanoid robots (G1, H2) and quadruped platforms (B2W). This ecosystem approach suggests long-term platform support, shared development tools, and interoperability—critical factors for anyone building robotics applications.
Unitree wasn't alone at the 2026 Spring Festival Gala. Other Chinese robotics companies including Galbot, Noetix, and MagicLab also featured robots in the broadcast, signaling a coordinated national effort to showcase domestic robotics capabilities.
China's government has identified humanoid robotics as a strategic technology priority, with provincial governments offering subsidies and incentives for robot manufacturers. The Spring Festival Gala appearance served dual purposes: entertaining domestic audiences while broadcasting China's robotics ambitions to the world.
For international buyers, this competitive landscape means more options, faster innovation, and—crucially—continued downward pressure on prices.
The humanoid robot kung fu performance at the 2026 Spring Festival Gala will be remembered as a watershed moment. Not because robots doing martial arts is inherently useful, but because it demonstrated capabilities that transfer directly to practical applications: dynamic balance, precise manipulation, real-time adaptation, and coordinated multi-robot operation.
Unitree has proven its robots can perform under pressure at the highest stakes imaginable. Now the question becomes: what will you build with one?
Whether you're a researcher, educator, or early commercial adopter, the Unitree G1 represents the most accessible entry point into humanoid robotics available today. Browse our complete selection of Unitree robots—including the G1, H2, and Go2 quadruped platforms—to find the right fit for your application.
Tesla is making its biggest strategic pivot since launching the Model S. Here's why ending production of its most iconic vehicles to manufacture humanoid robots signals a seismic shift in the company's identity—and validates the entire humanoid robotics industry.
After more than fourteen years of production, Tesla is pulling the plug on the Model S. The company announced on its Q4 2025 earnings call that both the Model S sedan and Model X SUV will cease production in Q2 2026 to free up manufacturing capacity at its Fremont, California factory—not for a new electric vehicle, but for Optimus humanoid robots.
"It's time to basically bring the Model S and X programs to an end with an honorable discharge, because we're really moving into a future that is based on autonomy," Elon Musk declared during the call. "That is slightly sad," he added, acknowledging the end of an era.
But sad or not, this represents one of the most dramatic strategic pivots in automotive history. Tesla is walking away from the vehicle that proved electric cars could work—the car that created Tesla's empire—to chase an unproven humanoid robot market where, by Musk's own admission, zero Optimus robots are currently doing "useful work" in Tesla's factories.
The Model S wasn't just any car—it was arguably the most important automobile of the 21st century. Before the Model S arrived in 2012, electric vehicles were slow, impractical, and appealing only to environmental guilt-trippers. Tesla's sedan changed everything.
The Model S pioneered over-the-air software updates, turning cars into upgradeable gadgets. It introduced Autopilot, laying the groundwork for autonomous driving technology. It turned Tesla into a tech company rather than just an automaker, with a stock price more reminiscent of Silicon Valley than Detroit.
But here's the cold reality: the Model S became irrelevant. More than 97% of Tesla's 418,227 vehicle deliveries in Q4 2025 were Model 3 and Model Y vehicles. The S, X, and Cybertruck combined accounted for fewer than 12,000 units—less than 3% of sales. In Tesla's financial reports, these once-flagship vehicles are now lumped under "Other Models."
Rather than continue pouring resources into declining luxury EVs, Tesla is converting those production lines for something Musk believes will be far bigger: robots that walk, talk, and work like humans.
The Fremont factory, Tesla's original production facility, is about to undergo its most significant transformation since Tesla acquired it from Toyota and GM in 2010.
According to Tesla's shareholder update, the company plans to unveil the Gen 3 version of Optimus in Q1 2026, featuring major upgrades including a new hand design. More importantly, this Gen 3 version is described as "the first design meant for mass production."
Tesla's stated goal is ambitious: production capacity of 1 million robots per year, with production starting before the end of 2026.
To put that in perspective, Tesla produced about 1.79 million vehicles globally in 2025. They're essentially building production capacity that could match half their entire vehicle output—but for robots.
"Because it is a completely new supply chain," Musk explained during the call, "there's really nothing from the existing supply chain that exists in Optimus." This means Tesla is building an entirely new manufacturing ecosystem from scratch.
Here's where the story gets complicated—and why investors and industry observers should approach Tesla's robotics claims with healthy skepticism.
On the same earnings call where Musk announced the factory conversion, he made a striking admission that directly contradicts years of Tesla's own claims.
"Well, we are still very much at the early stages of Optimus. It's still in the R&D phase," Musk said. "We have had Optimus do some basic tasks in the factory. But as we iterate on new versions of Optimus, we deprecate the old versions. It's not in usage in our factories in a material way. It's more so that the robot can learn."
Let's walk through what Tesla has said previously:
Now, one year later, the number doing useful work is zero. When asked during the earnings call how many Optimus robots Tesla actually has, Musk didn't answer the question.
This pattern—making bold near-term predictions that go unfulfilled—is why analysts at Electrek note they're "bullish on humanoid robots" but don't "really trust Musk leading this effort with this real credibility problem."
Despite the credibility concerns, Tesla's decision to end production of its most iconic vehicles sends an unmistakable signal to the market: humanoid robotics is real, and the biggest players are betting billions on it.
Consider what Tesla is actually doing:
When a company worth $800+ billion makes this kind of all-in strategic pivot, it validates the fundamental thesis that humanoid robots represent a massive market opportunity. Morgan Stanley projects the global humanoid robot market could reach $38 billion by 2035 and $5 trillion by 2050.
Dan Ives of Wedbush Securities calls Tesla "the best physical AI company in the world" and predicts Tesla could reach a $2 trillion market cap by end of 2026 based primarily on FSD and robotics growth—a 25% stock increase from current levels.
For the humanoid robotics industry as a whole, Tesla's pivot is a legitimizing event comparable to Apple entering the smartphone market. Even if Tesla stumbles on execution, their commitment signals to investors, suppliers, and talent that this market is worth pursuing.
While Tesla restructures for its robot ambitions, China has already established commanding market dominance.
Nearly 90% of all humanoid robots sold globally in 2025 were Chinese. Six of the highest-selling companies in the sector came from China.
Here are the 2025 unit sales according to market research firm Omdia:
That's right: two Chinese companies each outsold Tesla's entire 2025 production target of 5,000 units—a target Tesla failed to meet.
Unitree isn't standing still. The company's CEO Wang Xingxing announced they're targeting 20,000 humanoid robot shipments in 2026—nearly four times their 2025 output. The company is also preparing for a mid-2026 IPO, which would provide additional capital for expansion.
At China's Spring Festival Gala on January 28, 2026, Unitree's robots performed martial arts routines, 3-meter aerial flips, and trampoline somersaults—demonstrating capabilities that put Tesla's awkward walking demos to shame. Their G1 humanoids performed the kung fu sequence without any human intervention at the backend.
For buyers comparing options today, the Tesla Optimus vs Unitree G1 comparison shows just how competitive the pricing landscape has become.
While Unitree targets volume, Figure AI is focusing on enterprise deployments. The BMW manufacturing partnership continues, and Figure's approach of "walking before running"—deploying robots in controlled industrial settings before consumer markets—may prove more prudent than Musk's ambitious consumer-robot vision.
Norwegian company 1X has opened preorders for its NEO humanoid robot, with first customer deliveries planned for 2026. NEO is specifically designed for home use, targeting everyday tasks in unstructured residential settings rather than factory floors. This could give 1X a first-mover advantage in the consumer segment that Musk has promised but not delivered.
China now has over 150 robotics companies actively developing humanoid robots, compared to roughly 20 in the United States. "China is very good at AI, very good at manufacturing, and will definitely be the toughest competition for Tesla," Musk acknowledged at Davos.
Based on Tesla's announcements and historical track record, here's a realistic assessment:
Tesla's Official Timeline:
Reality Check:
Tesla promised 10,000 robots by end of 2025 and likely produced a few hundred. The company has yet to demonstrate an Optimus doing sustained, useful work without teleoperation (human remote control). Multiple supply chain reports throughout 2025 indicated Tesla's Optimus program was "in shambles," with the head of the program departing and production being delayed.
Analyst consensus suggests meaningful commercial production is more likely 2027-2028, with consumer-ready units arriving in late 2028 at earliest.
If you're a consumer interested in purchasing a humanoid robot, Tesla's pivot actually complicates your timeline:
The Good News:
The Bad News:
Our Recommendation:
If you're dead-set on a humanoid robot for home use, watch the 1X NEO closely—they're the most credible consumer play with actual delivery dates. For those willing to wait for Tesla, temper expectations: plan for 2028-2029 for a genuinely useful consumer product, not 2027.
For a comprehensive comparison of all available options, see our best humanoid robots guide and pricing breakdown.
Tesla's decision to end Model S and Model X production represents more than retiring two car models. It's a fundamental reorientation of a company that changed the automotive industry, now betting it can change the robotics industry too.
The Model S proved something that's now easy to take for granted: EVs can work, and ordinary people might actually want one. Now Tesla is attempting to prove something far more uncertain: that humanoid robots can work, and ordinary people (or at least ordinary factories) might actually want them.
Whether this pivot succeeds depends on whether Tesla can:
But even if Tesla stumbles, their commitment has permanently changed the industry's trajectory. When the world's most valuable automaker abandons its flagship vehicles to build robots, it signals to every investor, entrepreneur, and engineer that the humanoid robotics market is no longer science fiction—it's an emerging industry worth betting on.
The most important car of the 21st century is gone. What replaces it will define not just Tesla's future, but potentially the future of work itself.
The Sanctuary AI Phoenix stands at 170 cm (5'7") tall, weighs 70 kg (155 lbs), and represents one of the most intellectually ambitious humanoid robot programs on the planet. While competitors like Tesla and Figure chase headlines with flashy demos, Sanctuary AI has quietly built something different: a general-purpose robot whose real breakthrough isn't in its legs or its speed — it's in its hands and its mind. Powered by the proprietary Carbon AI system and equipped with 21-degree-of-freedom hydraulic hands that sense pressure down to 5 millinewtons, Phoenix is engineered to think and manipulate objects the way humans do. But with no public pricing, a prototype-phase status, and leadership upheaval in late 2024, is Sanctuary AI Phoenix worth the attention? This comprehensive Sanctuary AI Phoenix review breaks down every spec, every capability, and every limitation — so you can decide for yourself.
The Sanctuary AI Phoenix — a general-purpose humanoid robot built for dexterous industrial work.
Let's address the elephant in the room: Sanctuary AI does not publicly disclose the Phoenix price. The company operates strictly on a contact-sales, enterprise-first model. There is no e-commerce checkout, no pre-order page, and no published MSRP.
Based on our analysis of comparable general-purpose humanoid platforms currently in pilot or limited deployment — and considering Phoenix's advanced hydraulic hand system, proprietary Carbon AI software, and enterprise-grade build — we estimate the Sanctuary AI Phoenix price falls somewhere between $100,000 and $250,000 per unit for early commercial deployments. This is consistent with pricing from competitors like Agility Digit (~$250,000 for pilot programs) and Apptronik Apollo (targeting sub-$50,000 at scale).
Sanctuary's Magna International partnership likely involves custom pricing structures tied to volume commitments, and the company has signaled that reducing bill-of-materials costs is a priority with each generation — Generation 8 specifically highlights manufacturing cost reductions.
Here's how Phoenix's estimated pricing compares to the broader humanoid robot market:
The value proposition for Phoenix isn't about being the cheapest humanoid on the market — it never will be. It's about being the most dexterous. If your operation requires a robot that can sort small parts, handle delicate components, or perform assembly tasks that demand near-human finger precision, the Sanctuary AI Phoenix price may be justified by the labor it replaces. For organizations evaluating humanoid robot costs, Phoenix sits firmly in the premium industrial tier.
Here's what separates the Sanctuary AI Phoenix from virtually every other humanoid robot on the market: Sanctuary isn't trying to build the fastest runner or the most acrobatic bipedal platform. They're building the most dexterous general-purpose worker. And that strategic choice defines every aspect of Phoenix's performance profile.
Phoenix's hydraulic hands are the single most impressive subsystem on the robot. Each hand features 21 degrees of freedom — more than any other commercially available humanoid hand system. For context, the human hand has approximately 27 DOF. Phoenix is getting remarkably close.
The hands use proprietary miniaturized hydraulic valves rather than the electric motors found in competing platforms like Tesla Optimus or Figure 02. Sanctuary chose hydraulics for three specific reasons:
The results speak for themselves. Sanctuary has demonstrated in-hand object reorientation under extreme disturbance — including a 500g unexpected load — making it the first commercial humanoid to achieve this feat. This capability is critical for real-world manufacturing, where parts don't always arrive in perfect orientation.
In February 2025, Sanctuary integrated a new generation of tactile sensors into Phoenix's finger pads. Each pad contains a 7-cell touch sensor array using micro-barometers — the same miniaturized pressure sensors found in smartphones, repurposed for robotic dexterity.
The sensitivity numbers are striking: Phoenix can detect forces as low as 5 millinewtons (mN). Human fingertip sensitivity sits around 3 mN. That means Phoenix's sense of touch is within 40% of human capability — far ahead of any competitor that relies solely on vision-based manipulation.
As Dr. Jeremy Fishel, Sanctuary's principal researcher, explained: "Without tactile sensing, robots depend on video to interact with their environment. With video alone, you don't know you've touched something until well after the collision has physically caused the object to move."
The tactile system enables three critical capabilities:
Phoenix walks at approximately 4.8 km/h (3 mph) — roughly average human walking pace. It does not run, and Sanctuary has not prioritized bipedal agility in the way that other humanoid platforms have. The body uses electric actuation for locomotion while reserving hydraulics for the hands.
Generation 8 improved the range of motion in the wrists, hands, and elbows while reducing overall weight. The payload capacity of 25 kg (55 lbs) is competitive with the industrial humanoid category, though not class-leading — the FDROBOT TLIBOT, for instance, handles 145 kg.
For Sanctuary's target use cases — sorting parts, handling components, performing assembly tasks — walking speed and heavy lifting are secondary to what the hands can do. This is a deliberate engineering trade-off, and one that makes strategic sense given their Magna automotive partnership.
If Phoenix's hands are the hardware differentiator, Carbon AI is the software one. Carbon is Sanctuary's proprietary cognitive architecture — and it's fundamentally different from the AI approaches used by most humanoid competitors.
Carbon isn't just a neural network or a large language model bolted onto a robot. It's a hybrid cognitive system that combines multiple AI paradigms:
This hybrid approach gives Carbon something most competing systems lack: explainability. When Phoenix makes a decision — reach for this part, grasp it this way, place it there — Carbon can explain why it chose that plan. In regulated manufacturing environments, this audit trail matters enormously.
One of Sanctuary's most significant claims is that Phoenix can automate new tasks in under 24 hours. While the specifics vary by task complexity, TechCrunch verified demonstrations of the seventh-generation Phoenix learning to sort objects by color and type in structured environments within this timeframe.
The learning pipeline works through a combination of teleoperation (human operators controlling the robot remotely to generate training data) and reinforcement learning in simulation. Sanctuary leverages NVIDIA Isaac Lab — an open-source robot learning framework built on Isaac Sim — to train thousands of simulated hands simultaneously, dramatically accelerating the learning process.
As Sanctuary's team noted: "Our hands have kinematics beyond human capability, which cannot be accessed using analogous teleoperation. Online reinforcement learning in a simulated environment allows the learning algorithms to fully leverage the hands' capabilities."
Carbon translates natural language instructions into physical actions. Rather than requiring programming expertise, operators can describe tasks in conversational language, and Carbon generates reasoning, task, and motion plans to execute them. This dramatically lowers the barrier to deployment — a factory floor supervisor doesn't need to be a roboticist to direct Phoenix.
Carbon includes built-in support for human-in-the-loop supervision and fleet management. Multiple Phoenix robots can be monitored and directed by a single human operator, with the system handling autonomous execution of routine tasks and flagging situations that require human judgment.
The teleoperation capability serves dual purposes: it's both a production mode (allowing skilled operators to handle complex tasks remotely) and a data collection mechanism (every teleoperated session generates training data that improves autonomous performance).
The Phoenix sensor suite has been significantly upgraded in Generation 8, with improvements focused on data capture quality — which directly feeds Carbon AI's learning pipeline.
Phoenix uses a combination of depth cameras and RGB vision cameras. Generation 8 brings improved field of view and resolution to both systems. While Sanctuary hasn't disclosed specific camera models or resolutions, the upgrade was designed to increase the fidelity of visual data available for AI training.
Unlike competitors such as the Unitree H1 (which uses 3D LiDAR for 360° perception) or Tesla Optimus (which leverages Tesla's vision-only FSD AI stack), Phoenix's visual system is optimized for close-range manipulation tasks rather than long-range navigation. The cameras need to see what the hands are doing with high precision, not map an entire warehouse.
Force-torque sensors throughout the arms and wrists provide continuous feedback on the forces being applied during manipulation. This data integrates with the tactile sensors in the fingertips to create a comprehensive picture of every physical interaction.
Generation 8 includes improvements to Phoenix's audio and video systems for enhanced person-robot interaction. While specific microphone specs aren't public, the audio system supports natural language communication with Carbon AI and provides situational awareness in noisy manufacturing environments.
A key Generation 8 upgrade is the improved telemetry system designed specifically for high-quality data capture. Every sensor reading, every motor position, every force measurement is recorded and transmitted for use in training Carbon AI models. This "data-first" design philosophy means every minute of Phoenix operation contributes to making future autonomous behavior more robust.
Phoenix's design philosophy prioritizes function over aesthetics, though Generation 6 introduced "a bolder color palette and elevated textures" according to Sanctuary. The robot presents a clean, industrial appearance appropriate for factory environments.
At 170 cm (5'7") and 70 kg (155 lbs), Phoenix is deliberately human-sized. This matters for industrial deployment: the robot fits through standard doorways, operates at standard workbench heights, and can use tools designed for human hands. The human-like proportions also facilitate teleoperation — when a human operator controls Phoenix remotely, the 1:1 mapping between human and robot body dimensions makes control more intuitive.
Sanctuary hasn't disclosed specific materials or IP ratings for Phoenix. However, the Generation 8 design was explicitly built with manufacturing in mind — with emphasis on reduced bill-of-materials costs and simplified assembly, making the robot faster to commission and build. For industrial customers evaluating long-term deployment, this manufacturing-focused design suggests Sanctuary is planning for scale production rather than one-off prototypes.
The hands deserve special mention in any design discussion. Sanctuary has built five generations of robotic hands using electromechanical, cable-based, pneumatic, and ultimately hydraulic approaches before arriving at the current design. The miniaturized hydraulic valves represent years of R&D distilled into a compact, powerful hand that can exert significant force while maintaining the control needed for delicate manipulation.
The hydraulic approach enables what Sanctuary calls "beyond human capability" kinematics — the hands can achieve configurations and movements that human hands physically cannot, which becomes accessible through reinforcement learning rather than teleoperation.
Sanctuary iterates rapidly. In 8 generations since 2022, Phoenix has seen:
This annual iteration cycle demonstrates a commitment to continuous improvement that many well-funded competitors haven't matched.
This is Phoenix's marquee use case, anchored by the strategic partnership with Magna International — one of the world's largest automotive suppliers, manufacturing and assembling vehicles for Mercedes, Jaguar, and BMW. Magna's factories involve precisely the kind of dexterous manipulation tasks that Phoenix is designed for: sorting small mechanical parts, handling wiring harnesses, performing sub-assembly operations. The partnership aims to mature Phoenix technology for challenging manufacturing environments while scaling production. If you're in automotive manufacturing evaluating humanoid robot applications, Phoenix is one of the strongest candidates for dexterous work.
Phoenix's tactile sensing and fine manipulation capabilities make it well-suited for distribution centers where items of varying sizes, shapes, and fragility need to be sorted and packed. The blind picking capability — grasping items when vision is occluded — is particularly valuable in bin-picking scenarios where items overlap. While Agility Digit is purpose-built for logistics locomotion, Phoenix offers superior manipulation for tasks requiring finesse rather than speed.
Sanctuary AI lists energy as a target sector. Phoenix's potential here lies in inspection and maintenance tasks that require human-like dexterity in environments that are hazardous for human workers — handling electrical components, manipulating valves and switches, performing visual and tactile inspections of equipment. The teleoperation capability is especially valuable in dangerous environments where a human operator can control the robot from a safe distance.
The "general-purpose" designation matters. Unlike single-purpose industrial robots that are programmed for one task and require expensive retooling, Phoenix can theoretically be redeployed to different tasks within 24 hours. For a factory dealing with high product mix and frequent line changeovers, this flexibility could justify the higher upfront cost compared to traditional automation. As Sanctuary frames it: "To be general-purpose, a robot needs to be able to do nearly any work task, the way you'd expect a person to."
Phoenix's combination of tactile sensing (5 mN sensitivity), depth cameras, and force-torque measurement creates a comprehensive inspection platform. The robot can detect surface defects through touch, measure dimensional accuracy visually, and verify assembly quality through force testing — all autonomously or through teleoperation.
The Sanctuary AI Phoenix operates in a competitive landscape that includes some of the best-funded technology companies in the world. Here's how it stacks up against its closest competitors:
Figure 02 has massive financial backing and a high-profile BMW factory partnership. But when it comes to pure hand dexterity and tactile capability, Phoenix is in a different league. Figure's Helix foundation model is impressive for generalized learning, but Sanctuary's Carbon AI with its hybrid reasoning approach offers something Figure can't: explainable decision-making. For applications where auditable AI reasoning is required (automotive safety-critical components, for example), Phoenix has a clear edge.
Read our full comparison: Tesla Optimus vs Sanctuary AI Phoenix
Tesla's Optimus has the ultimate advantage: Tesla's manufacturing infrastructure and Elon Musk's stated goal of producing millions of units at $20,000-$30,000 each. If Tesla achieves this — and that's a significant "if" — Phoenix can't compete on price. But Phoenix isn't trying to. Sanctuary is targeting the high-value dexterous manipulation niche that Tesla's current hand design can't match. If your factory needs a robot that can handle small, fragile components with near-human touch sensitivity, Tesla Optimus isn't there yet. Phoenix is.
Understanding Phoenix requires understanding Sanctuary AI. Founded in 2018 in Vancouver, Canada, Sanctuary's founding team has a pedigree that reads like a who's-who of Canadian tech innovation:
The company has raised over $140 million in total funding from investors including Accenture Ventures, BDC Capital, InBC Investment, Magna International, BCE, Verizon Ventures, Workday Ventures, and a $30 million Strategic Innovation Fund contribution from the Government of Canada.
In November 2024, co-founder and CEO Geordie Rose was removed by the board. CTO Suzanne Gildert had already departed in April 2024. James Wells, previously the Chief Commercial Officer, stepped in as interim CEO. While leadership changes always introduce uncertainty, Wells brings commercial pragmatism to a company that had been primarily driven by its scientific vision. For potential customers, this shift may actually be positive — Wells' commercial background suggests a focus on getting Phoenix into paying customers' facilities rather than pursuing ever-more-ambitious research goals.
Morgan Stanley's Research division ranked Sanctuary AI third globally for published U.S. patents in humanoid robotics and embodied AI. This is significant — in a field where many companies are racing to file patents, Sanctuary's IP portfolio provides a defensive moat around its core hand dexterity and Carbon AI innovations.
The Sanctuary AI Phoenix price is not publicly disclosed. Sanctuary operates exclusively on a contact-sales model for enterprise customers. Based on our analysis of comparable industrial humanoid platforms and the advanced nature of Phoenix's hydraulic hand system, we estimate the price falls in the $100,000 to $250,000 range per unit. Organizations interested in Phoenix should contact Sanctuary AI directly through their official website to discuss pricing and pilot programs. For a broader view of humanoid robot pricing, see our humanoid robot cost guide.
Phoenix's primary differentiator is its industry-leading dexterous hand system. With 21 degrees of freedom per hand, hydraulic actuation, and tactile sensors sensitive to 5 millinewtons, Phoenix's hands are the most capable in any commercial humanoid program. While competitors focus on locomotion or general AI capabilities, Sanctuary has bet on manipulation as the key to general-purpose work — and the Magna International automotive partnership validates this approach.
No, Phoenix is not available for general purchase. The robot is currently in pilot deployment phase, available exclusively through enterprise partnership agreements. Sanctuary AI's primary commercial relationship is with Magna International for automotive manufacturing applications. The company has deployed earlier generations commercially and is expanding its customer base across automotive, manufacturing, and logistics sectors.
Carbon AI is Sanctuary's proprietary cognitive architecture — the "brain" that controls Phoenix. Unlike single-paradigm AI systems, Carbon combines symbolic reasoning, large language models, deep learning, and reinforcement learning into a unified system. This hybrid approach enables Phoenix to understand natural language instructions, plan task execution, control fine motor movements, and provide explainable reasoning for its decisions. Carbon also supports teleoperation and fleet management capabilities.
Yes. Sanctuary claims Phoenix can automate new tasks in under 24 hours through a combination of teleoperation (human-guided demonstration) and reinforcement learning. The company uses NVIDIA Isaac Lab to simulate training environments, allowing thousands of virtual hands to practice simultaneously. This sim-to-real transfer approach accelerates learning while reducing the risk of damaging physical hardware during training.
Phoenix and Tesla Optimus target different market segments despite both being "general-purpose" humanoids. Tesla aims for mass production at $20,000-$30,000 — a price point Phoenix will likely never match. However, Phoenix offers significantly more advanced hand dexterity (21 DOF hydraulic vs. Tesla's electric hands) and near-human tactile sensitivity. For high-value manufacturing tasks requiring fine manipulation, Phoenix is the superior choice. For mass-market general-purpose applications, Tesla's scale advantage may eventually prevail. See our detailed comparison.
Sanctuary AI is headquartered in Vancouver, British Columbia, Canada. The company was founded in 2018 and has operations primarily in North America, with customers and investors across Canada, the United States, Japan, and other countries.
For the right buyer, yes — with caveats. If you're an automotive manufacturer, logistics operator, or industrial facility with dexterous manipulation needs that can't be met by traditional automation, Phoenix offers capabilities no other humanoid can match. However, the lack of public pricing, the prototype-phase status, and recent leadership transitions mean you're buying into an early-stage platform. We recommend requesting a pilot deployment through Sanctuary AI to validate Phoenix's capabilities in your specific environment before committing to a larger rollout.
The Sanctuary AI Phoenix is the most dexterous humanoid robot you can evaluate today. Full stop. No other commercially available platform offers 21-DOF hydraulic hands with 5 mN tactile sensitivity, a hybrid cognitive architecture with explainable reasoning, and the ability to learn new manipulation tasks in under 24 hours. For organizations whose operations depend on fine manipulation — automotive assembly, electronics manufacturing, pharmaceutical packaging, precision logistics — Phoenix addresses a capability gap that no amount of Tesla hype or Figure funding has yet closed.
But Phoenix isn't for everyone. If you need a mass-market general-purpose humanoid at an accessible price point, wait for Tesla Optimus or look at 1X NEO. If you need a proven warehouse logistics solution today, Agility Digit is further along in commercial deployment. And if you're a researcher looking for an open SDK platform, Sanctuary's proprietary Carbon AI system may feel limiting compared to ROS-compatible alternatives like the Unitree G1.
The biggest risks with Sanctuary AI are financial and organizational, not technical. With ~$140M in funding against competitors with billions, and a recent leadership upheaval, the question isn't whether Phoenix can do the job — it's whether Sanctuary AI as a company can survive long enough to scale it. The Magna partnership and strong IP portfolio provide some insulation, but potential buyers should factor company risk into their evaluation alongside the impressive technical specs.
Ready to explore the Sanctuary AI Phoenix? View the full Sanctuary AI Phoenix listing on Robozaps or browse all humanoid robots for sale.
Last updated: February 1, 2026. Specs sourced from Sanctuary AI official documentation, press releases, TechCrunch, The Robot Report, and PR Newswire. Cross-referenced with the Robozaps robot database. Robozaps is a humanoid robot marketplace — we maintain hands-on product databases and may earn referral fees from qualifying purchases.
The best humanoid robot in 2026 is the Figure 03, followed by Tesla Optimus Gen 3 and Agility Robotics Digit. For budget buyers, the Unitree G1 at $13,500 offers the best value. The cheapest humanoid is Unitree's new R1 at $5,900. This expert-ranked guide covers all 28 major humanoid robots with verified specs, real pricing, and availability status.
Last updated: February 3, 2026 | 28 robots ranked by real-world deployment, capability, and value
The humanoid robot industry hit an inflection point in early 2026. Tesla is ramping Optimus Gen 3 production at its facilities. Boston Dynamics' electric Atlas shipped to Hyundai's Georgia Metaplant for real factory work. Figure AI's BotQ facility is tooled to produce 12,000 Figure 03 units annually. 1X Technologies started delivering NEO home robots to early adopters at $20,000. CES 2026 brought a wave of new entrants — Unitree's full-size H2 at $29,900, NEURA Robotics' Porsche-designed 4NE1 from €19,999, and LG's CLOiD home robot showcasing real household task demos.
This isn't hype anymore — it's hardware shipping. In this definitive guide, updated for February 2026, we rank and review 28 major humanoid robots available or in active deployment, complete with verified specs, real pricing, availability status, and use cases. Whether you're a buyer, investor, researcher, or simply tracking the future of robotics, this is the most comprehensive humanoid robot ranking on the internet.
Category Winners: Best Overall: Figure 03 | Best Value: Unitree G1 | Cheapest Humanoid: Unitree R1 ($5,900) | Best for Warehouses: Digit | Best for Healthcare: Fourier GR-2 | Best for Home: 1X NEO | Most Agile: Atlas (Electric) | Best Interaction: Ameca | Best Payload: Apollo & GR-2 | Most Affordable Full-Size: Kepler Forerunner
We evaluate every humanoid robot across five equally weighted criteria:
Robots working in real factories, warehouses, and hospitals always rank higher than those still in prototype or limited-pilot stages. We verify specs against manufacturer data sheets and cross-reference pricing with industry contacts. Last updated: February 1, 2026.
Manufacturer: Figure AI (Sunnyvale, CA) | Founded: 2022 | Funding: $1.9B+ (backed by Microsoft, OpenAI, NVIDIA, Jeff Bezos)
Figure AI's third-generation humanoid robot represents the most significant leap in commercial humanoid robotics to date. Released in October 2025, Figure 03 features a completely redesigned body with natural human proportions, the smoothest locomotion of any production humanoid, and an upgraded AI stack built on the company's proprietary Helix platform — enabling real-time speech, multi-step task reasoning, and autonomous error correction.
What sets Figure 03 apart is the combination of embedded palm cameras for precision manipulation, wireless charging capability, and visuomotor neural networks that deliver high frame rates with low latency. It's already performing real tasks in BMW's Spartanburg plant and other automotive facilities. Figure AI's new BotQ manufacturing facility is tooled to produce 12,000 units per year, with a stated target of 100,000 Figure 03 robots over the next four years. CEO Brett Adcock has said the company aims for full home autonomy by late 2026, with select home beta testers expected soon.
Key Specs:
Price: ~$130,000 (pilot program pricing) | View on Robozaps
Availability: Active pilot deployments with BMW and other automotive/tech manufacturers. BotQ facility ramping production. Commercial orders open for 2026.
Best For: Manufacturing assembly, logistics, quality inspection
Pros: Most complete AI + hardware package; real factory deployments; BotQ mass manufacturing; palm cameras for precision; strongest investor backing in industry
Cons: Not yet available for general purchase; limited track record vs. Digit in logistics; pricing still prohibitive for SMBs
Manufacturer: Tesla (Austin, TX) | Valuation context: Tesla's robotics division valued at up to $1T by some analysts
Tesla's Optimus robot made its biggest leap yet in January 2026. The company officially commenced mass production of Optimus Gen 3 at its Fremont, California factory — the same facility where Model S and Model X were built before Tesla discontinued those vehicles to make room for robot manufacturing. Musk has called this "the definitive start of the Physical AI era."
Gen 3 Optimus features redesigned actuators, improved 22-DoF hands, and Tesla's proprietary FSD-derived neural network trained on millions of hours of real-world factory data. The robots are already performing autonomous tasks inside Tesla's Austin Gigafactory and Fremont plant — including battery cell sorting, parts handling, box moving, and quality checks. Optimus Gen 3 has demonstrated smooth bipedal running, autonomous office navigation, and multi-step task execution.
Elon Musk confirmed in January 2026 that Tesla targets limited external sales by end of 2027, with a long-term consumer price target under $20,000. The Fremont line is designed for 1 million units per year capacity. If Tesla achieves this, Optimus could single-handedly make humanoid robots a mass-market product.
Key Specs:
Price: ~$25,000–$30,000 (estimated initial commercial price); long-term target under $20,000 | View on Robozaps
Availability: Limited internal production ongoing. External sales targeted for 2027+. Internal deployment at Tesla factories. Limited external sales expected end of 2027.
Best For: Factory automation, repetitive assembly, future home assistance
Pros: Mass production underway; unbeatable price-to-capability ratio at scale; Tesla's manufacturing expertise; massive AI training data; 1M unit/year capacity target
Cons: Not yet available for external purchase; Musk timelines historically optimistic; limited third-party validation
Manufacturer: Agility Robotics (Corvallis, OR) | Funding: $641M+ | Key partner: Amazon
Digit remains the gold standard for warehouse humanoid robots. With an industry-leading 8-hour battery life and a purpose-built design for logistics operations, Digit is already deployed in Amazon fulfillment centers and GXO facilities. Its adaptive grippers and AI-driven navigation let it handle diverse objects and environments with minimal human supervision.
Agility's "RoboFab" factory in Salem, Oregon — one of the first mass-production facilities dedicated to humanoid robots — has capacity to produce thousands of Digit units annually. This manufacturing maturity gives Digit a deployment advantage that most competitors can't match.
Key Specs:
Price: ~$250,000 (pilot and deployment pricing) | View on Robozaps
Availability: Commercially available. Active deployment with Amazon, GXO, and major logistics companies.
Best For: Warehouse picking/packing, truck loading/unloading, logistics
Pros: Best-in-class battery life; proven at scale with Amazon; dedicated manufacturing facility; most real-world deployment hours of any humanoid
Cons: High price point; limited dexterity compared to Figure 03; narrow focus on logistics tasks
Manufacturer: Boston Dynamics (Waltham, MA, subsidiary of Hyundai) | Heritage: 30+ years of bipedal robotics R&D
Boston Dynamics retired its iconic hydraulic Atlas in April 2024 and unveiled the all-electric Atlas — a fifth-generation humanoid built for real industrial work. The electric Atlas features 360-degree joint rotation at multiple points, a superior strength-to-weight ratio, and the most advanced sensor array of any humanoid: LiDAR, stereo cameras, RGB cameras, and depth sensors working in concert.
At CES 2026 in January, Hyundai showcased "Production Atlas" performing autonomous parts sequencing in a mock factory — identifying heavy car components with its advanced AI reasoning system and precisely placing them onto assembly lines. The robot's torso spun 180 degrees while its legs stayed planted, demonstrating capabilities unconstrained by human biology. Hyundai announced Atlas is now deployed at its Georgia Metaplant, moving from R&D project to capital equipment. This makes Atlas the most expensive — but arguably most capable — humanoid robot in actual commercial production use.
Key Specs:
Price: ~$420,000 (enterprise only)
Availability: Shipping to Hyundai Georgia Metaplant. Enterprise deployments expanding 2026.
Best For: Automotive manufacturing, heavy industrial tasks, R&D, hazardous environments
Pros: Most mechanically capable humanoid ever; 360° joint rotation; now in actual production deployment; decades of R&D heritage
Cons: Extremely expensive (~$420K); enterprise-only; heavy for its height; limited production capacity
Manufacturer: Unitree Robotics (Hangzhou, China) | Funding: $150M+ Series B
The Unitree G1 shattered expectations by delivering a genuinely capable humanoid robot at a price point that puts it within reach of researchers, educators, startups, and enthusiasts. Starting at just $13,500, the G1 offers up to 43 degrees of freedom (in the EDU configuration), 3D LiDAR, depth cameras, and dexterous hands capable of complex manipulation tasks like opening bottles, soldering, and folding laundry.
The G1 uses reinforcement learning to continuously improve its motor skills, and Unitree's strong developer community provides extensive open-source tools and tutorials. It's the most accessible entry point into humanoid robotics by a wide margin — though Unitree's new R1 (see #16) aims to undercut it at just $5,900.
Key Specs:
Price: Starting at $13,500 (base); ~$21,600 (standard); ~$27,000 (EDU with 43 DoF) | View on Robozaps
Availability: ️ Unverified for purchase now — ships worldwide.
Best For: Research, education, AI training, development platform, hobbyists
Pros: Unbeatable price; ships worldwide today; strong developer community; up to 43 DoF; ROS2 compatible; continuous OTA updates
Cons: Small stature limits real-world industrial use; short battery life (2 hrs); limited payload (3 kg)
Manufacturer: Sanctuary AI (Vancouver, Canada) | Key partners: Magna International, Microsoft
Sanctuary AI's Phoenix is purpose-built for general-purpose work with an emphasis on dexterous manipulation. Now in its eighth generation, Phoenix features the industry's most advanced tactile sensors in its hands, controlled by Sanctuary's proprietary Carbon™ AI system — the company's bid to create "the world's first human-like intelligence in a general-purpose robot."
Carbon™ enables Phoenix to learn new tasks faster than any competing system — Sanctuary claims 88% reduction in task training time from Gen 7 to Gen 8. Phoenix is being piloted in retail, automotive manufacturing (with Magna), and logistics environments.
Key Specs:
Price: ~$40,000 (estimated) | View on Robozaps
Availability: Pilot deployments expanding in 2026. Partnerships with Magna and Microsoft.
Best For: Retail, logistics, manufacturing, general-purpose labor
Pros: Fastest task-learning AI; excellent dexterity; strong price point; partnerships with major companies
Cons: Not yet broadly commercially available; less proven at scale than Digit or Figure 03
Manufacturer: Apptronik (Austin, TX) | Funding: $403M Series A (backed by B Capital, Capital Factory, Google)
Apollo is the workhorse of the humanoid world. With the highest payload capacity in its class (55 lbs / 25 kg), a modular design, hot-swappable batteries, and built-in safety features including LED displays and force control, Apollo is designed for the most physically demanding industrial environments. Apptronik's NASA collaboration heritage and Google operations testing add serious credibility.
Apollo is active in pilot programs with Mercedes-Benz for automotive manufacturing and with logistics companies for warehouse operations. The company targets a sub-$50,000 price point for mass deployment — which would make it one of the most affordable full-size industrial humanoids.
Key Specs:
Price: Sub-$50,000 target for mass deployment | View on Robozaps
Availability: Pilot programs with Mercedes-Benz, Google, and logistics firms.
Best For: Heavy lifting, warehouse operations, manufacturing, construction assistance
Pros: Highest payload capacity; hot-swappable batteries; strong safety features; NASA heritage; Mercedes-Benz + Google partnerships
Cons: Final pricing unconfirmed; enterprise-only; limited AI sophistication compared to Figure 03 or Phoenix
Manufacturer: 1X Technologies (Sunnyvale, CA / Oslo, Norway) | Backed by: OpenAI, Samsung, EQT Ventures
NEO is the world's first humanoid robot truly purpose-built for the home — and it's no longer just a concept. 1X Technologies has begun delivering NEO to early adopters in the US in 2026, making it the first consumer humanoid robot to actually ship. Its lightweight design (just 66 lbs / 30 kg), home-safe soft actuators, and emphasis on natural human interaction make it fundamentally different from industrial humanoids.
At $20,000 (or $499/month subscription), NEO uses teleoperation to train its AI initially, with fully autonomous operation planned for later iterations. Available in 3 colors (Tan, Gray, Dark Brown), NEO can run at up to 22 km/h and receives monthly AI software updates. Privacy-first design includes face-blurring cameras and user-defined no-go zones.
Key Specs:
Price: $20,000 (or $499/month subscription) | View on Robozaps
Availability: Shipping to early adopters in the US. Preorders open.
Best For: Home assistance, elder care, smart home integration, companionship
Pros: First consumer humanoid actually shipping; affordable; OpenAI AI backing; subscription option; privacy-first design
Cons: Initially teleoperated (1X operators can see through cameras); US-only; first-gen product — expect early adopter issues
Manufacturer: Unitree Robotics (Hangzhou, China)
The H1-2 is Unitree's upgraded full-size humanoid — a significant improvement over the original H1 with added arm dexterity (7 DoF per arm vs. 4), ankle articulation (2 DoF vs. 1), and a more robust 70 kg frame. It was the first full-size humanoid in China capable of running at up to 13 km/h, and at ~$90,000, it bridges the gap between affordable research platforms and expensive industrial humanoids.
Unitree's M107 joint motors deliver peak torque density of 189 N.m/kg — claimed to be the highest in the world. The H1-2 supports 3D LiDAR, depth cameras, ROS2 compatibility, and continuous OTA software updates.
Key Specs:
Price: ~$90,000 | View on Robozaps
Availability: Available for purchase. Ships globally.
Best For: Research, light assembly, locomotion studies, public demonstrations
Pros: Best value full-size humanoid; world-record walking speed; 7-DoF arms; replaceable battery; strong developer ecosystem
Cons: Limited manipulation capability vs. dedicated industrial robots; Chinese-only documentation for some features
Manufacturer: Fourier Intelligence (Shanghai, China) | Heritage: Leading rehabilitation robotics company
Building on the GR-1's foundation, the GR-2 represents Fourier's evolved humanoid platform with 53 degrees of freedom, improved dexterity, and a taller 175 cm frame. Fourier's unique advantage is its rehabilitation robotics heritage — the company already deploys exoskeletons and therapy robots in 40+ countries, giving GR-2 an unmatched pathway into healthcare environments. Mass production is targeting 2026.
Key Specs:
Price: ~$150,000 (projected) | View on Robozaps
Availability: Pilot deployments in healthcare and industrial settings. Mass production planned 2026.
Best For: Physical therapy, rehabilitation, elder care, heavy industrial tasks
Pros: Best payload-to-weight ratio; built by rehab robotics experts; 53 DoF; global distribution in healthcare
Cons: Not yet mass-produced; less AI sophistication than Figure 03 or Phoenix
Manufacturer: UBTECH Robotics (Shenzhen, China) | Public company: Listed on HKEX (9880)
Walker S1 is a manufacturing powerhouse with 41 servo joints and large language model integration. Already deployed at Audi's China plant for quality inspection and at NIO's electric vehicle factory, Walker S1 was the first humanoid to demonstrate multi-robot collaboration in a real factory setting. UBTECH's partnership with Foxconn to explore iPhone assembly marks another major milestone.
Key Specs:
Price: Enterprise pricing (contact manufacturer) | View on Robozaps
Availability: Commercially available. Deployed at Audi China and NIO.
Best For: Quality inspection, assembly line support, manufacturing
Pros: Proven factory deployments; publicly traded (stability); LLM integration; first multi-humanoid collaboration
Cons: Enterprise pricing opaque; primarily China-focused; slow walking speed (3 km/h)
Manufacturer: RobotEra (Beijing, China)
The RobotEra STAR1 burst onto the scene as one of the fastest and most agile Chinese humanoids. Standing 171 cm tall, it reaches speeds of 4 m/s (14.4 km/h) — making it the fastest walking humanoid robot in production — and features 12-DoF dexterous hands. Its competitive pricing at ~$96,000 positions it as a strong mid-range option.
Key Specs:
Price: ~$96,000
Availability: Orders open for 2026 delivery.
Best For: Logistics, service deployments, dynamic environments requiring speed
Pros: Fastest humanoid walking speed; competitive pricing; dexterous 12-DoF hands
Cons: Newcomer with limited deployment track record; smaller ecosystem than Unitree
Manufacturer: Stardust Intelligence / Astribot (Shenzhen, China)
Astribot S1 stunned the robotics world with demo videos showing it performing tasks with speed and precision exceeding human capabilities — pouring liquids, ironing clothes, flipping objects, and writing calligraphy with fluid motion. S1's 52 degrees of freedom and AI-driven upper-body dexterity are genuinely impressive, with arm end-effector speeds up to 10 m/s.
Key Specs:
Price: ~$80,000 (estimated) | View on Robozaps
Availability: Pilot deployments in China. Broader availability expected 2026.
Best For: Dexterous manipulation, service tasks, food preparation, light manufacturing
Pros: Exceptional upper-body dexterity; fast arm speed; competitive pricing
Cons: Demo-to-reality gap unclear; limited deployments; newer company
Manufacturer: AgiBot (Shanghai, China, incubated by Shanghai AI Lab)
AgiBot A2 excels in service environments where human-like interaction matters. With AI-powered sensors and an ergonomic design, it can perform precision tasks like threading a needle while engaging customers in natural conversation. Mass production started in December 2024 with 962+ units already produced — positioning it among high-volume humanoid manufacturers. Manufacturer claims certification for China, US, and European markets.
Key Specs:
Price: Contact manufacturer | View on Robozaps
Availability: Available. Mass production active with 962+ units shipped.
Best For: Customer service, exhibitions, marketing events, guided tours
Pros: Mass production underway; triple-certified; strong conversational AI; precision manipulation
Cons: China-focused availability; enterprise pricing not transparent
️ Note: Manufacturer website unavailable at time of verification. Specs are based on industry reports and may not reflect current product status.

Manufacturer: Kepler Robotics (Shanghai, China)
Kepler's Forerunner humanoid targets the sweet spot between affordability and industrial capability. With 40 degrees of freedom, a full-size 178 cm frame, and an estimated price point around $30,000, Kepler is positioning itself as the affordable industrial humanoid for factories that can't justify $100K+ robots.
Key Specs:
Price: ~$30,000 (estimated) | View on Robozaps
Availability: ️ Unverified programs active. Broader availability expected mid-2026.
Best For: Light manufacturing, assembly, inspections, service tasks
Pros: Extremely competitive price for full-size humanoid; 40 DoF; good battery life
Cons: Early-stage company; limited deployment data; heavier than competitors
Manufacturer: Unitree Robotics (Hangzhou, China)
The Unitree R1 is a game-changer: at just $5,900, it's the cheapest humanoid robot ever offered. Unveiled in late 2025 and now available for pre-order, the R1 is an ultra-lightweight 25 kg bipedal robot targeting the consumer and education markets. From the same company that proved affordable humanoids are possible with the G1, the R1 pushes accessibility to a new level.
While specifications are still limited compared to the G1 or H1-2, the R1 represents a psychological price breakthrough — a full humanoid robot for less than a used car. It's an entry point for schools, hobbyists, and early adopters who want to experience bipedal robotics without a $13,500+ investment.
Key Specs:
Price: $4,900–$5,900
Availability: Pre-order open. Shipping expected 2026.
Best For: Education, hobbyists, entry-level robotics, entertainment
Pros: Cheapest humanoid robot ever; ultra-lightweight; from established manufacturer (Unitree); bipedal walking
Cons: Limited specs publicly available; likely limited autonomous capabilities; pre-order only; very compact form factor
Manufacturer: Unitree Robotics (Hangzhou, China)
Unveiled at CES 2026 and immediately available for pre-order, the Unitree H2 bridges the gap between the compact G1 and the research-grade H1. At $29,900, it's the cheapest full-size (180 cm) humanoid robot ever offered. Featuring 31 degrees of freedom, a lifelike face with expression capability, depth perception, and quick-swap batteries, the H2 targets both commercial service and educational markets. Available in Commercial ($29,900) and EDU variants.
Key Specs:
Price: $29,900 (Commercial) | View on Robozaps
Availability: Pre-order open. Shipping expected April 2026.
Best For: Commercial service, education, enterprise pilots, robotics development
Pros: Cheapest full-size humanoid ever; 31 DoF; lifelike expressions; from proven manufacturer; quick-swap batteries
Cons: Not yet shipping; limited real-world deployment data; new platform
Manufacturer: NEURA Robotics (Metzingen, Germany)
The 4NE1 Gen 3.5 is the first humanoid robot designed in collaboration with Studio F.A. Porsche. Unveiled at CES 2026 with pre-orders now open, the flagship model costs €98,000 while the smaller 4NE1 Mini starts at just €19,999 — making it one of the most affordable full humanoids from a Western manufacturer. Features include patented artificial skin for proximity detection, 100 kg lifting capacity, the Neuraverse OS for fleet-wide skill sharing, and NVIDIA Isaac GR00T-powered multimodal reasoning.
Key Specs:
Price: €19,999 (Mini) / €98,000 (Gen 3.5) — pre-orders open with €100 refundable deposit
Availability: Pre-order open. Deliveries expected 2026.
Best For: Industrial automation, domestic assistance, fleet deployments
Pros: Exceptional lifting capacity (100kg); Porsche design pedigree; fleet skill-sharing; artificial safety skin; affordable Mini variant
Cons: Not yet shipping; German pricing (€); relatively new to humanoid market

Manufacturer: LG Electronics (Seoul, South Korea)
Debuted at CES 2026 as the centerpiece of LG's "Zero Labor Home" vision, CLOiD is a home humanoid robot that was demonstrated performing real household tasks — folding laundry, loading dishwashers, and preparing food. Unlike bipedal designs, CLOiD uses a wheeled base with a height-adjustable torso, dual 7-DoF arms, and five-fingered hands for fine manipulation. Powered by LG's "Affectionate Intelligence" and a Vision-Language-Action model, it integrates deeply with LG's ThinQ smart home ecosystem.
Key Specs:
Price: Not yet announced
Availability: Prototype demonstrated at CES 2026. Production timeline TBD.
Best For: Home assistance, smart home integration, elderly care
Pros: Backed by LG's massive manufacturing; real household task demos; ThinQ ecosystem integration; height-adjustable design
Cons: Not commercially available; wheeled (no bipedal); no pricing; prototype stage
Manufacturer: Xiaomi (Beijing, China)
CyberOne is Xiaomi's first humanoid robot, featuring emotion detection via computer vision, 21 degrees of freedom, and the full weight of Xiaomi's hardware engineering ecosystem. Still primarily a research platform, but Xiaomi's massive manufacturing infrastructure means CyberOne could scale rapidly if the technology matures.
Key Specs:
Price: ~$105,000 (estimated R&D cost; not commercially available) | View on Robozaps
Availability: R&D prototype. Not available for purchase.
Best For: Research, companion robotics R&D
Pros: Backed by tech giant; emotion recognition; lightweight
Cons: Very limited payload (1.5 kg); not commercially available; only 21 DoF
Manufacturer: Engineered Arts (Falmouth, UK)
Ameca is the world's most expressive humanoid robot, built for human interaction, research, and entertainment. Its hyper-realistic facial expressions, conversational AI with GPT integration, and lifelike gestures make it unmatched for customer-facing roles, exhibition demos, and HRI research. The Tritium OS platform enables embodied AI development. Deployed in schools, elder care, museums, and trade shows worldwide.
Key Specs:
Price: $100,000–$140,000 (depending on configuration)
Availability: Available for purchase and lease.
Best For: Human interaction research, exhibitions, hospitality, education
Pros: Unmatched expressiveness; GPT-powered conversation; proven in customer-facing environments
Cons: Cannot walk; mostly stationary; limited physical task capability

Manufacturer: XPENG Robotics (Guangzhou, China)
XPENG's IRON humanoid brings automotive engineering precision to humanoid robotics. With an industry-leading 200 degrees of freedom, 22-DoF hands, a solid-state battery, and 720° vision system, IRON achieves remarkably natural movement. Powered by XPENG's Turing AI / VLA 2.0 platform, it's partnered with Baosteel for industrial monitoring. The sheer DOF count is unprecedented — making IRON one of the most biomechanically advanced humanoids in development.
Key Specs:
Price: Not yet announced | View on Robozaps
Availability: Prototype. Baosteel industrial partnership active.
Best For: Industrial inspection, guided tours, equipment monitoring
Pros: Most degrees of freedom of any humanoid (200); solid-state battery; XPENG's manufacturing scale; 22-DoF hands
Cons: Not commercially available; prototype stage; no pricing announced
Manufacturer: 1X Technologies (Sunnyvale, CA / Oslo, Norway)
EVE holds the distinction of being one of the first AI-powered humanoid robots to enter the commercial workforce. Using a wheeled base for stability, EVE features strong grippers, panoramic vision cameras, and custom AI that learns and improves from experience. Deployed in security, manufacturing support, and logistics.
Key Specs:
Price: Enterprise pricing (contact manufacturer)
Availability: Commercially available for enterprise deployment.
Best For: Security, manufacturing support, logistics
Pros: Proven workforce deployment; reliable wheeled mobility; learning AI; long battery life
Cons: Wheeled, not bipedal; enterprise-only pricing

Manufacturer: Humanoid Ltd (UK)
The HMND 01 Alpha is the UK's first humanoid robot designed for industrial use — and it was built in a remarkable 7 months. Standing an imposing 220 cm tall (7'3"), it's the tallest humanoid robot on this list. Available in both wheeled and bipedal variants, it moves at 7.2 km/h and carries 15 kg payloads. The KinetIQ AI framework provides vision, manipulation, navigation, and reasoning capabilities.
Key Specs:
Price: Contact sales
Availability: Available. Built and shipping from UK.
Best For: Industrial automation, manufacturing, logistics
Pros: Tallest humanoid (220cm); fast development cycle; available now; wheeled + bipedal options
Cons: New company with limited track record; limited ecosystem

Manufacturer: Fauna Robotics (USA)
Fauna Sprout takes a different approach to home humanoids — it's a lightweight, interactive home robot built as an open developer platform. At $50,000, it sits between consumer and enterprise pricing, targeting developers, researchers, and tech-forward homes. Early customers include Disney, Boston Dynamics, UC San Diego, and NYU — a strong signal that Sprout has serious technical credibility despite being from a young company.
Key Specs:
Price: $50,000
Availability: Available for purchase.
Best For: Home R&D, developer platform, research institutions
Pros: Strong early customer list; developer-friendly; home-safe design
Cons: Expensive for consumers; limited public specs; new company
Manufacturer: SoftBank Robotics (Tokyo, Japan)
Though no longer in mass production, Pepper remains the most widely deployed service humanoid in history. Over 27,000 units have been sold and thousands continue operating in banks, airports, hospitals, and retail stores worldwide.
Key Specs:
Price: Previously ~$1,800/month; now special order programs
Availability: Discontinued for mass sales; special orders and refurbished available.
Best For: Customer greeting, retail assistance, education
Pros: Most proven track record (27,000+ units); 12-hour battery; multilingual
Cons: No longer in production; outdated AI vs. 2026 competitors
Manufacturer: SoftBank Robotics / Aldebaran (Paris, France)
NAO is the world's most popular educational humanoid robot. Standing just 58 cm tall, this bipedal robot speaks 20 languages, features 25 degrees of freedom, and is used in thousands of schools, universities, and research labs. At ~$9,000, it's the most accessible bipedal humanoid for educational institutions.
Key Specs:
Price: ~$9,000
Availability: Available for purchase.
Best For: Education, autism therapy research, programming instruction
Pros: Most deployed educational robot; multilingual; affordable; extensive curriculum
Cons: Very small; minimal physical capability; aging hardware
Manufacturer: Promobot (Philadelphia, PA / Perm, Russia)
Promobot V.4 is the most customizable service humanoid available — hotel concierge, museum guide, medical assistant, or security system. With facial recognition, document scanning, payment processing, and natural language conversation, over 800 units operate in 47 countries.
Key Specs:
Price: $25,000–$50,000
Availability: Commercially available in 47 countries.
Best For: Hotel concierge, museum tours, healthcare intake
Pros: Highly customizable; proven in 47 countries; 800+ units; integrated payments
Cons: Wheeled, not bipedal; limited physical capability; less advanced AI than 2026 competitors
Factory & Manufacturing: Figure 03 offers the best AI + dexterity combination. Tesla Optimus Gen 3 will be the value leader once externally available. Walker S1 and Atlas are proven in automotive plants. For heavy parts, Apollo's 25 kg payload leads the field.
Warehouse & Logistics: Digit is the undisputed leader — 8-hour battery, Amazon-proven, mass-manufactured. RobotEra STAR1 offers speed advantage at a lower price. Apollo handles the heaviest loads.
Healthcare & Rehabilitation: Fourier GR-2 is purpose-built by rehabilitation robotics experts with 50 kg payload for patient support. No other humanoid comes close in this vertical.
Research & Education: Unitree G1 at $13,500 is unbeatable for labs. NAO at $9,000 for K-12 education. H1-2 at $90,000 for full-size research. The new Unitree R1 at $5,900 is the cheapest entry point ever.
Customer Service & Hospitality: Ameca for maximum wow-factor. Promobot V.4 for practical concierge tasks. AgiBot A2 for AI-native conversation.
Home & Personal Use: 1X NEO ($20,000 or $499/month) is the first purpose-built home humanoid now shipping. Fauna Sprout ($50K) for developer-minded homes. Tesla Optimus is the long-term home robot play, but 2+ years away from consumers.
Under $10,000: Unitree R1 ($5,900) — cheapest humanoid ever. SoftBank NAO (~$9,000) — educational only.
$10,000–$25,000: Unitree G1 ($13,500–$27,000), 1X NEO ($20,000), Promobot V.4 ($25,000+).
$25,000–$100,000: Unitree H2 ($29,900), Tesla Optimus (~$25K–$30K est.), Kepler Forerunner (~$30K est.), Phoenix (~$40K), Fauna Sprout ($50K), Astribot S1 (~$80K), H1-2 ($90K), RobotEra STAR1 (~$96K).
$100,000–$250,000: Figure 03 (~$130K), Ameca ($100K–$140K), Fourier GR-2 (~$150K), Digit (~$250K).
$250,000+: Boston Dynamics Atlas (~$420,000) — enterprise-only, premium capabilities.
The humanoid robotics market is experiencing explosive growth. Valued at $2.03 billion in 2024, it's projected to surpass $13 billion by 2029 according to MarketsandMarkets — a nearly 7x increase in five years. Several forces are driving this transformation:
January 2026 marked the true beginning of humanoid mass production. Tesla commenced Optimus Gen 3 manufacturing at Fremont with a 1M unit/year capacity target. Figure AI's BotQ facility is tooled for 12,000 Figure 03 units per year. Agility's RoboFab produces thousands of Digits annually. AgiBot has shipped 5,000+ A2 units globally. China's Eyou opened the world's first automated production line for humanoid robot joints. This supply chain maturation will drive prices down 30–50% over the next 2–3 years.
Every top humanoid robot in 2026 runs on advanced AI — vision-language models for understanding commands and environments, large language models for natural conversation, and reinforcement learning for physical tasks. Figure 03's Helix platform can hold conversations while performing multi-step assembly. Tesla Optimus leverages FSD neural networks. Sanctuary's Carbon™ cuts task training time by 88%. This AI integration is what separates today's humanoids from the clunky automatons of five years ago.
BMW (Figure), Hyundai (Atlas), Audi (Walker S1), Mercedes-Benz (Apollo), NIO (Walker S1), Baosteel (XPENG IRON), and Foxconn (UBTECH) are integrating humanoid robots into their factories. Tesla discontinued Model S and X to make room for Optimus production at Fremont. The automotive industry's adoption signals that humanoid robots are transitioning from novelty to necessity.
In 2023, the cheapest capable humanoid was around $13,500 (Unitree G1). In 2026, Unitree's R1 hit $5,900 and 1X's NEO subscription is just $499/month. Kepler targets $30K for a full-size industrial humanoid. Tesla targets sub-$20K at scale. Within 3–5 years, expect capable humanoids under $5,000 — approaching appliance pricing.
Chinese companies (Unitree, AgiBot, RobotEra, Fourier, UBTECH, Kepler, Astribot, XPENG, EngineAI) now produce more humanoid robot models than any other country. The Chinese government has formed industrial coalitions supporting humanoid development. Meanwhile, the US leads in AI sophistication (Figure, Tesla, Boston Dynamics, 1X, Apptronik) and venture capital. For buyers, this competition means more options, lower prices, and faster innovation.
2026 marks the first time humanoid robots are actually shipping to homes. 1X's NEO is delivering to early adopters at $20,000 (or $499/month). Fauna Sprout offers a developer platform at $50K. Figure 03 is targeting home betas. Tesla targets sub-$20,000 consumer Optimus by 2028. The home humanoid era that science fiction promised is beginning now.
If you're looking for the best humanoid robot for sale, here are your options:
The Figure 03 ranks as the best overall humanoid robot in 2026, combining advanced AI (Helix platform), 48+ degrees of freedom, dexterous palm-camera manipulation, real-world factory deployments with BMW, and BotQ mass manufacturing. For specific use cases: Digit leads in logistics, Unitree G1 in affordability, Fourier GR-2 in healthcare, and NEO for home use.
Humanoid robot prices in 2026 range from $5,900 (Unitree R1) to over $420,000 (Boston Dynamics Atlas). Most commercial humanoids fall in the $20,000–$250,000 range. The cheapest capable humanoids: Unitree R1 ($5,900), Unitree G1 ($13,500), 1X NEO ($20,000 or $499/mo). Tesla's Optimus targets under $20,000 long-term.
Yes — for the first time, home humanoid robots are actually shipping. 1X Technologies' NEO is delivering to early adopters at $20,000 (or $499/month) and is designed specifically for home use. The Unitree G1 ($13,500) is affordable for enthusiasts. Fauna Sprout ($50K) serves developer-minded homes. Tesla Optimus may become the ultimate home robot once it reaches consumer pricing (expected 2028+).
The Unitree R1 at just $5,900 is the cheapest humanoid robot ever offered — now available for pre-order. For a more capable option, the Unitree G1 at $13,500 offers up to 43 degrees of freedom, 3D LiDAR, and ships worldwide. The SoftBank NAO at ~$9,000 is a small educational robot, not a full-size humanoid.
For wheeled humanoids: SoftBank Pepper leads at ~12 hours. For service robots: Promobot V.4 at 8+ hours. For bipedal humanoids: Agility Robotics Digit is the endurance champion at 8 hours of continuous bipedal operation — crucial for warehouse shifts.
Today's best humanoid robots can: pick and pack warehouse orders (Digit), perform factory assembly and quality inspection (Figure 03, Walker S1, Atlas), navigate stairs and uneven terrain (Atlas, H1-2), hold natural conversations (Ameca, Phoenix), assist with physical therapy (GR-2), carry up to 55 lbs (Apollo, GR-2), run at up to 22 km/h (NEO), and operate up to 8 hours on a charge (Digit). They cannot yet reliably cook complex meals, drive vehicles, or fully replace human judgment in unstructured environments.
Not replacing — augmenting. In 2026, humanoid robots handle repetitive, physically demanding, or dangerous tasks that are difficult to staff. The US manufacturing labor shortage exceeds 500,000 unfilled positions. Tesla literally couldn't find enough humans to run its factories, which partly drove the Optimus program. The World Economic Forum estimates automation will create more new jobs in robot maintenance, programming, and oversight than it eliminates.
The XPENG IRON leads by a massive margin with 200 degrees of freedom, thanks to its biomimetic muscle and joint system. The Fourier GR-2 follows with 53 DoF, and Astribot S1 features 52 DoF.
Industry leaders predict humanoid robots could be widespread in homes by the early 2030s. 1X's NEO is already shipping at $20,000. Tesla targets sub-$20,000 Optimus by 2028, with millions of units by 2029. Unitree's R1 at $5,900 shows prices are dropping fast. More conservative estimates suggest mainstream adoption (>10% of households) by 2035, once prices drop below $5,000 and AI supports unsupervised operation.
Bipedal humanoid robots (Atlas, Figure 03, Digit) walk on two legs, enabling stairs, uneven terrain, and human-designed spaces. Mechanically more complex with shorter battery life. Wheeled humanoids (Pepper, EVE, Promobot) are more energy-efficient and stable but can't handle stairs or rough terrain. The best choice depends on your environment — warehouses with multiple floors need bipedal; flat retail spaces work great with wheeled.
The 28 best humanoid robots of 2026 represent a genuine inflection point in technology history. Tesla is mass-producing Optimus Gen 3 at Fremont. Atlas is shipping to Hyundai factories. Figure 03's BotQ is ramping to 12,000 units per year. NEO is delivering to homes. And the cheapest humanoid robot now costs just $5,900.
Prices range from $5,900 to $420,000, with the sweet spot rapidly moving downward. AI capabilities are advancing at breakneck speed — each generation dramatically more capable than the last. With China and the US racing to lead the humanoid revolution, innovation is accelerating on every front.
Whether you're evaluating humanoid robots for your business, researching investment opportunities, or tracking the future of technology, 2026 is the year these machines proved they belong. The question is no longer "will humanoid robots work?" — it's "which one is right for you?"
Stay ahead of the humanoid revolution. Bookmark this page — we update our rankings monthly as new robots launch and existing ones evolve. For individual robot reviews, pricing, and buying advice, explore more on blog.robozaps.com and browse humanoid robots for sale on Robozaps.
The Unitree G1 is the best humanoid robot most people can actually buy in 2026. At $13,500–$27,000, it offers 23–43 degrees of freedom, 3D LiDAR, depth cameras, and dexterous manipulation — making it ideal for research, education, and development. For home use, the 1X NEO at $20,000 is now shipping to early adopters. Enterprise buyers should consider Agility Digit for warehouse logistics or Figure 03 for manufacturing.
Humanoid robot prices range from $5,900 to over $400,000 depending on capability and use case. Budget-friendly options include Unitree R1 ($5,900), Unitree G1 ($13,500+), and 1X NEO ($20,000). Mid-range industrial robots like Apollo and Phoenix cost $40,000–$150,000. Premium robots like Boston Dynamics Atlas ($420,000) and Digit ($250,000) target enterprise deployments with proven reliability.
Not yet. As of February 2026, Tesla has not opened pre-orders or sales for Optimus. Mass production of Optimus Gen 3 began at the Fremont factory in January 2026, but these units are for Tesla's internal use. Elon Musk targets limited external sales by late 2027 at $20,000–$30,000. There is no waitlist — be wary of any third-party site claiming to accept Tesla robot pre-orders.
The Unitree R1 at $5,900 is the cheapest humanoid robot announced for 2026, currently in pre-order. The most affordable full-capability humanoid available now is the Unitree G1 starting at $13,500. For education, the SoftBank NAO at ~$9,000 is a smaller 58cm robot widely used in schools and research.
The 1X NEO is currently the best humanoid robot designed specifically for home use. At $20,000, it features a lightweight 30kg body, quiet operation, and AI trained for household tasks like tidying, fetching items, and basic chores. It's now shipping to early adopters. Tesla's Optimus also targets home use but won't be available until late 2027 at earliest. LG's CLOiD home robot was announced at CES 2026 but has no pricing or availability yet.
In 2026, humanoid robots can reliably perform: warehouse logistics (Digit moves boxes at Amazon), manufacturing assembly (Atlas works at Hyundai, Figure 03 at BMW), quality inspection (Walker S1 deployed in factories), and basic home tasks (NEO handles simple chores). They can walk, climb stairs, manipulate objects, respond to voice commands, and learn new tasks through demonstration. Full autonomous home assistance — cooking, cleaning, childcare — remains limited and experimental.
Match the robot to your use case: Research/Education → Unitree G1 ($16K) or NAO ($9K). Warehouse/Logistics → Agility Digit or Apptronik Apollo. Manufacturing → Figure 03 or Boston Dynamics Atlas. Home/Personal → 1X NEO or wait for Tesla Optimus. Entertainment/Exhibitions → Ameca. Consider availability (can you buy it now?), price, support ecosystem, and whether you need RaaS (Robot-as-a-Service) vs. outright purchase.
Last updated: February 3, 2026 | Pricing and availability verified against manufacturer sources, CES 2026 announcements, and industry contacts.
1X NEO ($20k) vs Unitree H2 ($29.9k) — complete specs comparison, pricing, safety features, and decision framework. See which 2026 home humanoid robot is right for you.
Two humanoid robots. Two very different philosophies. One shared goal: bringing automation into your home. The 1X NEO ($20,000) and Unitree H2 ($29,900) represent the first generation of humanoid robots actually priced for consumer pre-order—and both are shipping in 2026.
If you've been waiting for the "buy now" button on a home humanoid, it's finally here. But which robot deserves your deposit? This comprehensive 1X NEO vs Unitree H2 comparison breaks down everything: price, size, capabilities, design philosophy, and which robot fits your specific needs.
| Specification | 1X NEO | Unitree H2 |
|---|---|---|
| Price | $20,000 (or $499/mo) | $29,900 |
| Deposit | $200 (refundable) | $2,500 |
| Height | Compact (approx. 5'6" / 167cm) | 5'11" / 180cm |
| Weight | ~30 kg (66 lbs) | 70 kg (154 lbs) |
| DOF | Not disclosed | 31 degrees of freedom |
| Body Type | Soft-bodied, tendon-driven | Rigid, industrial-grade |
| Target Use | Home consumers only | Home + Commercial |
| Delivery | 2026 (US) | April 2026 (North America) |
| Software | OTA updates, basic autonomy | OTA updates, SDK available |
| NA Distributor | Direct from 1X | ToborLife (exclusive) |
| Color Options | Tan, Gray, Dark Brown | White/Gray standard |
1X Technologies (formerly Halodi Robotics) designed the NEO from the ground up for one thing: living safely with humans. Unlike every other humanoid robot on the market, NEO uses a soft-bodied, tendon-driven design inspired by human muscle and ligament structure.
Where most humanoids are rigid metal skeletons with exposed joints and hard surfaces, NEO is soft. Its "muscles" are artificial tendons that give it compliant, human-like movement. If you bump into NEO—or if it bumps into you, your furniture, or your kids—it yields like a person would, not like a metal cabinet falling over.
This isn't just marketing. It's the fundamental engineering decision that makes NEO viable for home use. Traditional industrial robots require safety cages. NEO can hand you a cup of coffee without risk of cracking your ribs.
1X offers two ways to own NEO:
The subscription model is interesting. At $499/month, you'd pay off the $20,000 purchase price in 40 months (3.3 years). If you're uncertain about the technology or expect rapid hardware improvements, subscribing makes sense. If you're committed long-term, buying saves money.
The deposit is only $200 and fully refundable—the lowest commitment of any humanoid robot on the market.
NEO ships with "basic autonomy"—1X's term for foundational home navigation and simple task completion. The company has been transparent that capabilities will expand significantly through over-the-air (OTA) software updates.
Expected launch capabilities include:
Unitree—the Chinese robotics company famous for the $13,500 G1 humanoid and ultra-affordable quadruped robots—launched the H2 at CES 2026 with a clear message: this is the cheapest full-size humanoid robot ever sold.
At 182cm (6'0") and 70kg (154 lbs), the H2 is built to human adult specifications. It can reach the same shelves you can, use the same tools you use, and operate in environments designed for people without modification.
The trade-off: H2 uses traditional rigid robotics construction. It's stronger and more capable than NEO, but also harder and less forgiving in collisions. This isn't a robot you want falling on your toddler.
The H2 packs serious hardware into its frame:
The H2 is priced at $29,900—approximately $10,000 more than NEO, but dramatically cheaper than any comparable full-size humanoid. The Unitree G1 (smaller model) starts at $13,500, making the H2's per-capability value remarkable.
Pre-orders require a $2,500 deposit through ToborLife, the exclusive North American distributor. Use code TOBORBOTINFO200 for $200 off.
Delivery is targeted for April 2026 in North America.
Winner: 1X NEO
At $20,000 vs $29,900, NEO costs roughly 33% less than H2. The subscription option ($499/month) drops the entry barrier even further. NEO's $200 refundable deposit vs H2's $2,500 makes testing the waters dramatically easier.
However, value depends on your use case. If you need a full-size humanoid's capabilities, the H2's $29,900 is still historic—full-size humanoids from Figure, Boston Dynamics, and others cost $50,000-$250,000+.
Winner: 1X NEO
NEO's soft-bodied design is purpose-built for homes with children, pets, and fragile furniture. The tendon-driven actuators provide inherent compliance—if something goes wrong, the robot yields rather than pushes through.
H2 is a traditional industrial robot. It's safe by industrial standards (which require extensive risk assessment), but it's not designed with the assumption that a 3-year-old might grab its leg mid-stride.
Winner: Unitree H2
The H2's larger frame, 31 DOF, and industrial actuators give it strictly superior physical capabilities. It can reach higher, carry more, and perform more complex movements than NEO's compact form allows.
If you need a robot that can help with garage work, reach top shelves, or handle heavier loads, H2 is the clear choice.
Winner: Unitree H2
NEO is designed exclusively for home consumers. H2 targets both home and light commercial applications. If you're a small business owner considering a robot for both home and shop, H2 can do double duty.
Unitree also offers an "H2 EDU" variant for educational institutions, signaling broader SDK and developer support.
Winner: Tie
Both robots promise OTA updates to expand capabilities over time. Unitree has a longer track record shipping consumer robotics (their Go1/Go2 quadrupeds have active developer communities), while 1X has focused more on commercial deployments with their EVE robot.
Neither company has published detailed SDK documentation for their consumer humanoids yet. This is a "wait and see" category.
Winner: Tie (slight edge to H2)
H2 has a specific target: April 2026 for North America. NEO's timeline is "2026 for US orders" without a specific month. Unitree's track record with the G1 launch suggests they can hit aggressive timelines; 1X is newer to consumer delivery.
Neither is objectively "better"—they serve different needs. NEO is safer and more affordable for home-only use. H2 is more capable and versatile for mixed home/commercial applications. If you have children, NEO's soft-bodied design is the responsible choice.
NEO costs $20,000 (or $499/month subscription) with a $200 deposit. H2 costs $29,900 with a $2,500 deposit. NEO is 33% cheaper with a 12x lower deposit requirement.
Both target 2026. H2 has a specific April 2026 target for North America. NEO says "2026 for US orders" without specifying a month.
Yes. NEO's $200 deposit is fully refundable. H2's $2,500 deposit terms vary—check with ToborLife for current cancellation policy.
NEO, by a significant margin. Its soft-bodied, tendon-driven design was specifically engineered for safe human coexistence. H2 uses traditional rigid construction that requires more caution around vulnerable family members.
Yes. Both companies commit to OTA (over-the-air) software updates that will expand capabilities post-purchase. This is the new standard for humanoid robots—you're buying a platform, not a fixed product.
For most home buyers, the 1X NEO is the better choice. Its safety-first design, lower price, minimal deposit, and subscription option make it the lower-risk entry point into home humanoid robotics. If you're buying a robot to help around the house and you have family members who might interact with it unpredictably, NEO is the responsible pick.
However, the Unitree H2 wins if you need serious capability. Its full adult-size frame, 31 DOF, and industrial-grade construction mean it can tackle tasks NEO simply cannot. If you're a power user, developer, or small business owner who needs maximum versatility, H2 delivers more robot for the premium.
Both robots represent a historic moment: the first humanoids priced for real consumers, shipping in 2026. The wait is almost over.
Ready to pre-order? See 1X NEO on Robozaps | See Unitree H2 on Robozaps | Compare all humanoid robots
Last updated: February 20, 2026. Specifications sourced from 1X Technologies and Unitree official announcements. Robozaps is a humanoid robot marketplace — we maintain comprehensive product databases and may earn referral fees from qualifying purchases.
Unitree G1 and H2 robots performed the world's first fully autonomous humanoid kung fu routine at China's 2026 Spring Festival Gala. With 20,000 units planned for 2026, here's what this means for the robotics industry.
On February 16, 2026, approximately 679 million people watched something unprecedented unfold on their screens: dozens of Unitree humanoid robots performing fully autonomous kung fu on the stage of China's Spring Festival Gala. No teleoperation. No pre-programmed dance moves. Just pure, AI-driven martial arts that included backflips, weapon handling, and a record-breaking 7.5-rotation Airflare spin.
This wasn't a tech demo in a sanitized laboratory. This was the Unitree Spring Festival 2026 moment—broadcast live to the largest television audience on Earth during China's equivalent of the Super Bowl.
The Spring Festival Gala (Chunwan) is China's most-watched annual broadcast, traditionally featuring music, dance, and cultural performances. In 2026, Unitree Robotics made history by debuting its Unitree G1 robot fleet alongside the larger H2 models in a segment titled "Cyber Real Kung Fu."
According to Unitree's official press release, this marked "the world's first fully autonomous humanoid robot cluster martial arts performance." The routine wasn't just impressive—it shattered multiple technical records:
The H2 models added dramatic flair, appearing in Monkey King armor and even riding Unitree's B2W quadruped robot dogs as "somersault clouds"—a reference to the legendary Chinese folk hero Sun Wukong.
The star of the show, the Unitree G1, represents Unitree's push into affordable humanoid robotics. Here are the key specifications that enabled those viral kung fu moves:
| Specification | Unitree G1 |
|---|---|
| Height | 127 cm (4.2 ft) |
| Weight | 35 kg (77 lbs) |
| Degrees of Freedom | 23+ (up to 43 with dexterous hands) |
| Max Walking Speed | 2+ m/s (over 7 km/h) |
| Battery Life | ~2 hours (quick-swap design) |
| Starting Price | $13,500 USD (base model) |
What sets the G1 apart isn't just hardware—it's the AI driving it. Unitree implemented systematic upgrades across algorithms, hardware, and systems specifically for the gala performance. The robots used reinforcement learning combined with force-position hybrid control, enabling the precise, fluid movements that captivated the global audience.
While the G1 handled the acrobatic kung fu sequences, Unitree's H2 model brought the theatrical presence. Standing taller and built for heavier industrial applications, the H2 appeared at both the Beijing main venue and the Yiwu sub-venue.
Priced at approximately $29,900, the H2 targets different use cases—warehouse logistics, manufacturing assistance, and heavy-duty manipulation tasks. Its appearance at the gala demonstrated that Unitree isn't just building research platforms; they're building a full product ecosystem for China humanoid robots 2026 and beyond.
Perhaps more significant than the viral performance was what Unitree founder Wang Xingxing announced afterward. In an interview with tech outlet 36Kr, Wang revealed that Unitree plans to ship between 10,000 and 20,000 humanoid robots in 2026.
To put this in perspective:
This isn't aspirational marketing—it's a signal that humanoid robots are transitioning from experimental technology to commercial products. When a company commits to shipping 20,000 units, supply chains, manufacturing processes, and quality control systems must already be in place.
If you're considering purchasing a humanoid robot—whether for research, education, or early commercial applications—the Unitree Spring Festival 2026 performance carries several implications:
Live performances don't lie. When robots execute complex martial arts routines autonomously in front of hundreds of millions of viewers, it validates the underlying technology in ways that controlled demos never can. The G1's performance proves it can handle dynamic, unpredictable scenarios—not just scripted laboratory tasks.
At $13,500 for the base G1, Unitree offers arguably the best value proposition in the humanoid market. Competitors like Boston Dynamics' Atlas remain research-only platforms without consumer pricing. Tesla's Optimus has yet to reach general availability. The G1 is shipping now.
Unitree's 20,000-unit production target means more robots in the field, more edge cases discovered, and faster iteration on reliability issues. Early adopters benefit from a company operating at scale rather than building one-off prototypes.
The gala showcased integration between Unitree's humanoid robots (G1, H2) and quadruped platforms (B2W). This ecosystem approach suggests long-term platform support, shared development tools, and interoperability—critical factors for anyone building robotics applications.
Unitree wasn't alone at the 2026 Spring Festival Gala. Other Chinese robotics companies including Galbot, Noetix, and MagicLab also featured robots in the broadcast, signaling a coordinated national effort to showcase domestic robotics capabilities.
China's government has identified humanoid robotics as a strategic technology priority, with provincial governments offering subsidies and incentives for robot manufacturers. The Spring Festival Gala appearance served dual purposes: entertaining domestic audiences while broadcasting China's robotics ambitions to the world.
For international buyers, this competitive landscape means more options, faster innovation, and—crucially—continued downward pressure on prices.
The humanoid robot kung fu performance at the 2026 Spring Festival Gala will be remembered as a watershed moment. Not because robots doing martial arts is inherently useful, but because it demonstrated capabilities that transfer directly to practical applications: dynamic balance, precise manipulation, real-time adaptation, and coordinated multi-robot operation.
Unitree has proven its robots can perform under pressure at the highest stakes imaginable. Now the question becomes: what will you build with one?
Whether you're a researcher, educator, or early commercial adopter, the Unitree G1 represents the most accessible entry point into humanoid robotics available today. Browse our complete selection of Unitree robots—including the G1, H2, and Go2 quadruped platforms—to find the right fit for your application.
Complete Figure AI review covering Figure 01, 02, 03 robots, the revolutionary Helix AI system, BMW deployment results, $39B valuation, funding from OpenAI/Microsoft/NVIDIA, pricing estimates, and what makes Figure a humanoid robotics leader.
Figure AI is one of the most ambitious humanoid robotics companies in the world. With backing from OpenAI, Microsoft, Jeff Bezos, NVIDIA, and Intel—plus a $39 billion valuation—they're redefining what humanoid autonomy looks like. Here's everything you need to know.
Figure AI has emerged as one of the most ambitious and well-funded humanoid robotics companies in the world. With backing from OpenAI, Microsoft, Jeff Bezos, NVIDIA, and Intel, plus a valuation that hit $39 billion in late 2025, Figure isn't just building robots—they're redefining what humanoid autonomy looks like.
In this comprehensive review, I'll break down everything you need to know about Figure AI: their robot lineup, the revolutionary Helix AI system, real-world deployment results at BMW, and whether Figure lives up to the hype.
| Aspect | Details |
|---|---|
| Founded | May 2022 by Brett Adcock |
| Headquarters | San Jose, California |
| Valuation | $39 billion (September 2025) |
| Total Funding | ~$1.7 billion+ |
| Key Investors | OpenAI, Microsoft, NVIDIA, Jeff Bezos, Amazon, Intel Capital |
| Robot Lineup | Figure 01 (retired), Figure 02 (commercial), Figure 03 (latest) |
| AI System | Helix / Helix 02 Vision-Language-Action (VLA) |
| Key Deployment | BMW Spartanburg plant (30,000+ vehicles produced) |
| Target Markets | Manufacturing, logistics, home assistance |
| Production Goal | 100,000 robots over next 4 years |
Figure AI was founded in May 2022 by Brett Adcock, a serial entrepreneur who previously founded Archer Aviation (eVTOL aircraft, NASDAQ: ACHR) and Vettery (acquired by Adecco for $100M+). Adcock assembled Figure's founding team from alumni of Boston Dynamics, Tesla, Google DeepMind, and Apple—a who's who of robotics and AI talent.
The company's mission is deceptively simple: give AI a physical body. While chatbots and large language models have transformed digital interactions, Figure believes the real transformation happens when AI can manipulate the physical world—folding laundry, loading dishwashers, assembling products on factory floors.
Figure's funding trajectory reflects extraordinary investor confidence in humanoid robotics:
| Round | Date | Amount | Valuation | Key Investors |
|---|---|---|---|---|
| Seed | 2022 | $70M | ~$300M | Parkway Venture Capital |
| Series A | May 2023 | $100M | ~$1B | Intel, Parkway |
| Series B | Feb 2024 | $675M | $2.6B | OpenAI, Microsoft, NVIDIA, Jeff Bezos, Amazon, Intel |
| Series C | Sept 2025 | $1B+ | $39B | Parkway, Brookfield, NVIDIA |
The February 2024 Series B was a watershed moment. Having OpenAI, Microsoft, and NVIDIA all invest in a robotics startup signaled that the biggest names in AI see humanoid robots as the next frontier. The partnership with OpenAI was particularly significant—it positioned Figure to leverage cutting-edge language models for robot reasoning.
Figure 01 was the company's proof-of-concept humanoid, unveiled on March 2, 2023 and deployed in limited testing through 2023-2024. Standing at 5'6" (168 cm) and weighing 132 lbs (60 kg), Figure 01 demonstrated that Figure could build a functional bipedal humanoid.
Key Figure 01 specs:
Figure 01's main purpose was learning—both for the company and for the AI models that would eventually become Helix. It was used for the initial BMW partnership testing and helped Figure understand what manufacturing environments actually demand from a humanoid.
Figure 02 represented Figure's first commercially viable humanoid. Announced in 2024 and deployed at BMW's Spartanburg plant, Figure 02 proved that humanoids could work real shifts in real factories.
| Specification | Figure 02 |
|---|---|
| Height | 5'6" (168 cm) |
| Weight | 155 lbs (70 kg) |
| Payload Capacity | 44-55 lbs (20-25 kg) |
| Battery | 2.25 kWh lithium-ion |
| Runtime | 5+ hours |
| Compute | 3x more powerful than Figure 01 |
| Cameras | 6 onboard cameras |
| Hand DoF | 16 degrees of freedom per hand |
| Walking Speed | 1.2 m/s |
The Figure 02's key innovations were:
After 11 months at BMW (1,250+ hours of runtime, 90,000+ parts loaded), Figure retired Figure 02 to make way for Figure 03. The lessons learned—particularly around forearm reliability and wrist electronics—directly shaped the next generation.
Figure 03, introduced in late 2025, represents Figure's most ambitious robot yet. It's not just an industrial workhorse—it's designed to eventually enter homes.
| Specification | Figure 03 |
|---|---|
| Height | ~5'6" (estimated, similar to F.02) |
| Weight | 9% lighter than Figure 02 |
| Camera System | 2x frame rate, 25% latency, 60% wider FOV |
| Palm Cameras | Embedded in each hand for in-hand visual feedback |
| Tactile Sensors | Fingertip sensors detecting forces as small as 3 grams |
| Hand Design | Softer, more compliant fingertips |
| Battery | UN38.3 certified, multi-layer safety protection |
| Charging | 2 kW wireless inductive charging via foot pads |
| Data Offload | 10 Gbps mmWave wireless |
| Covering | Soft textiles (washable, replaceable) |
What makes Figure 03 special:
Hardware matters, but Helix is what makes Figure's robots genuinely intelligent. Helix is a Vision-Language-Action (VLA) model—a neural network that directly converts visual input and language commands into robot actions.
Traditional robot programming works like this:
Helix works differently:
This is transformative. Instead of programming thousands of individual behaviors, Figure can simply tell the robot what to do.
Helix uses a "dual-system" architecture inspired by cognitive psychology:
System 2 (S2): The Slow Thinker
System 1 (S1): The Fast Reactor
This separation is elegant: S2 can "think" about what to do while S1 handles the split-second motor control needed for smooth movements. It's similar to how humans consciously decide to pick up a cup (slow, deliberate) while the actual reaching-and-grasping happens automatically (fast, reactive).
With Helix, Figure robots can:
The "pick up anything" capability is particularly impressive. Helix learned from only ~500 hours of demonstration data—a fraction of what other VLA systems require—yet generalizes to thousands of novel objects.
Helix 02, unveiled on January 27, 2026, extended the original Helix from upper-body control to full-body control. This is a massive leap.
The New System 0 Layer
Helix 02 adds a third layer to the architecture:
| System | Role | Speed | Parameters |
|---|---|---|---|
| System 2 | Scene understanding, language | 7-9 Hz | 7B |
| System 1 | Full-body joint targets | 200 Hz | 80M |
| System 0 | Whole-body balance & coordination | 1,000 Hz | 10M |
System 0 is trained on 1,000+ hours of human motion data and handles the physics of staying upright while moving and manipulating. It replaced 109,504 lines of hand-engineered C++ code with a single neural network.
The Dishwasher Demonstration
To showcase Helix 02, Figure released a 4-minute continuous task video: a humanoid robot autonomously unloading a dishwasher, walking across a kitchen, placing items in cabinets, reloading the dishwasher, and starting it.
Key stats from this demonstration:
Figure called this "the longest horizon, most complex task completed autonomously by a humanoid robot to date." Whether or not that's strictly true, it's undeniably impressive—especially the seamless integration of walking, reaching, balancing, and fine manipulation.
New Dexterity Tasks
Helix 02's palm cameras and tactile sensors enable tasks that were impossible with vision alone:
Talk is cheap. The real test of any industrial robot is whether it can survive a factory floor. Figure's 11-month deployment at BMW's Spartanburg plant provides hard data.
| Metric | Result |
|---|---|
| Parts Loaded | 90,000+ |
| Runtime Hours | 1,250+ |
| Vehicles Contributed To | 30,000+ BMW X3s |
| Estimated Robot Steps | 1.2+ million |
| Distance Walked | 200+ miles |
The task had strict requirements:
The challenge: placing three sheet-metal parts within 5 mm tolerance in just 2 seconds—while moving fast enough to keep up with the line.
The BMW deployment wasn't just about proving capability—it generated invaluable data for Figure 03:
BMW hasn't announced plans to deploy Figure 03 yet, but the partnership validated Figure's approach to humanoid manufacturing.
Figure robots are not available for consumer purchase. As of February 2026:
While Figure hasn't published official pricing, industry estimates suggest:
| Model | Estimated Price | Target Market |
|---|---|---|
| Figure 02 | $30,000-$50,000 | Commercial/industrial |
| Figure 03 | $50,000-$100,000 (speculative) | Commercial, eventually home |
For context, Tesla's Optimus is targeting ~$25,000-$30,000, while Unitree's G1 starts at $16,000. Figure is positioning higher on capability rather than competing purely on price.
Figure is currently focused on:
CEO Brett Adcock has stated the goal is to have Figure 03 "in select homes" by late 2026, but this will likely be limited pilot programs rather than broad consumer availability.
How does Figure stack up against other humanoid players?
| Company | Robot | Height | Weight | Price Est. | Key Strength |
|---|---|---|---|---|---|
| Figure | Figure 03 | 5'6" | ~140 lbs | $50-100K | Helix AI, full-body autonomy |
| Tesla | Optimus Gen 2 | 5'8" | 127 lbs | $25-30K | Scale, cost, Tesla ecosystem |
| 1X Technologies | NEO | 5'5" | 66 lbs | ~$30K | Lightweight, home-focused |
| Agility Robotics | Digit | 5'9" | 141 lbs | ~$200K+ | Logistics-optimized |
| Unitree | G1 | 4'3" | 77 lbs | $16K | Affordable, research-friendly |
| Boston Dynamics | Atlas | 4'11" | 196 lbs | Not for sale | Most athletic movements |
Figure's advantages:
Figure's challenges:
Figure hasn't released official pricing. Industry estimates suggest Figure 02 was $30,000-$50,000 for commercial deployments, and Figure 03 may be $50,000-$100,000. These robots are not currently available for consumer purchase.
Not yet. Figure 03 is designed with home environments in mind (soft materials, wireless charging, safety features), but consumer availability isn't expected until late 2026 at earliest—and even then, it will likely be limited pilot programs.
Helix is Figure's proprietary Vision-Language-Action (VLA) model. It allows Figure robots to understand natural language commands ("Pick up the red cup") and translate them directly into motor actions. Helix 02, released January 2026, extended this to full-body control including walking and balance.
Figure retired the Figure 02 fleet in November 2025 after the BMW deployment concluded and Figure 03 entered production. The lessons learned from Figure 02's 1,250+ hours of factory runtime directly informed Figure 03's design improvements.
Tesla Optimus is targeting lower price (~$25,000-$30,000) and higher volume (millions of units eventually). Figure is pursuing higher capability with Helix AI and has proven factory deployment. Tesla has more manufacturing scale; Figure has more sophisticated AI integration.
No. Figure AI is a private company. It has raised over $1.7 billion in venture funding at a $39 billion valuation (September 2025). There's no announced timeline for an IPO.
Based on demonstrated capabilities:
Key investors include OpenAI, Microsoft, NVIDIA, Jeff Bezos (personal investment), Amazon, Intel Capital, Parkway Venture Capital, and Brookfield Asset Management.
Figure AI is building exactly what the humanoid robot industry needs: capable hardware paired with genuinely intelligent AI. The Helix system—especially Helix 02's full-body autonomy—represents the most sophisticated integration of language understanding and physical manipulation I've seen in a commercial humanoid.
The BMW deployment proves Figure isn't just making demo videos. 90,000+ parts loaded, 30,000+ vehicles contributed to, zero-reset shifts—that's real work.
But let's be clear: Figure 03 isn't ready for your living room yet. The dishwasher demos are impressive but carefully controlled. Real homes are chaotic, unpredictable, and full of edge cases that will test any AI system.
Who should care about Figure AI right now?
Who should wait?
Figure has the funding, the talent, the AI, and the manufacturing roadmap. The question isn't whether humanoid robots will work in factories and homes—it's how fast Figure can scale. With 100,000 robots targeted over four years and a $39 billion valuation backing them, Figure AI is one of the most serious bets in robotics.
For individual robot reviews, see our Figure 01 Review, Figure 02 Review, and Figure 03 Review.
Tesla is ending Model S/X production in Q2 2026 to convert Fremont factory for Optimus humanoid robot manufacturing. While Musk admits zero robots are doing useful work today, this pivot validates the humanoid market—even as Unitree targets 20,000 units and Chinese manufacturers control 90% market share.
Tesla is making its biggest strategic pivot since launching the Model S. Here's why ending production of its most iconic vehicles to manufacture humanoid robots signals a seismic shift in the company's identity—and validates the entire humanoid robotics industry.
After more than fourteen years of production, Tesla is pulling the plug on the Model S. The company announced on its Q4 2025 earnings call that both the Model S sedan and Model X SUV will cease production in Q2 2026 to free up manufacturing capacity at its Fremont, California factory—not for a new electric vehicle, but for Optimus humanoid robots.
"It's time to basically bring the Model S and X programs to an end with an honorable discharge, because we're really moving into a future that is based on autonomy," Elon Musk declared during the call. "That is slightly sad," he added, acknowledging the end of an era.
But sad or not, this represents one of the most dramatic strategic pivots in automotive history. Tesla is walking away from the vehicle that proved electric cars could work—the car that created Tesla's empire—to chase an unproven humanoid robot market where, by Musk's own admission, zero Optimus robots are currently doing "useful work" in Tesla's factories.
The Model S wasn't just any car—it was arguably the most important automobile of the 21st century. Before the Model S arrived in 2012, electric vehicles were slow, impractical, and appealing only to environmental guilt-trippers. Tesla's sedan changed everything.
The Model S pioneered over-the-air software updates, turning cars into upgradeable gadgets. It introduced Autopilot, laying the groundwork for autonomous driving technology. It turned Tesla into a tech company rather than just an automaker, with a stock price more reminiscent of Silicon Valley than Detroit.
But here's the cold reality: the Model S became irrelevant. More than 97% of Tesla's 418,227 vehicle deliveries in Q4 2025 were Model 3 and Model Y vehicles. The S, X, and Cybertruck combined accounted for fewer than 12,000 units—less than 3% of sales. In Tesla's financial reports, these once-flagship vehicles are now lumped under "Other Models."
Rather than continue pouring resources into declining luxury EVs, Tesla is converting those production lines for something Musk believes will be far bigger: robots that walk, talk, and work like humans.
The Fremont factory, Tesla's original production facility, is about to undergo its most significant transformation since Tesla acquired it from Toyota and GM in 2010.
According to Tesla's shareholder update, the company plans to unveil the Gen 3 version of Optimus in Q1 2026, featuring major upgrades including a new hand design. More importantly, this Gen 3 version is described as "the first design meant for mass production."
Tesla's stated goal is ambitious: production capacity of 1 million robots per year, with production starting before the end of 2026.
To put that in perspective, Tesla produced about 1.79 million vehicles globally in 2025. They're essentially building production capacity that could match half their entire vehicle output—but for robots.
"Because it is a completely new supply chain," Musk explained during the call, "there's really nothing from the existing supply chain that exists in Optimus." This means Tesla is building an entirely new manufacturing ecosystem from scratch.
Here's where the story gets complicated—and why investors and industry observers should approach Tesla's robotics claims with healthy skepticism.
On the same earnings call where Musk announced the factory conversion, he made a striking admission that directly contradicts years of Tesla's own claims.
"Well, we are still very much at the early stages of Optimus. It's still in the R&D phase," Musk said. "We have had Optimus do some basic tasks in the factory. But as we iterate on new versions of Optimus, we deprecate the old versions. It's not in usage in our factories in a material way. It's more so that the robot can learn."
Let's walk through what Tesla has said previously:
Now, one year later, the number doing useful work is zero. When asked during the earnings call how many Optimus robots Tesla actually has, Musk didn't answer the question.
This pattern—making bold near-term predictions that go unfulfilled—is why analysts at Electrek note they're "bullish on humanoid robots" but don't "really trust Musk leading this effort with this real credibility problem."
Despite the credibility concerns, Tesla's decision to end production of its most iconic vehicles sends an unmistakable signal to the market: humanoid robotics is real, and the biggest players are betting billions on it.
Consider what Tesla is actually doing:
When a company worth $800+ billion makes this kind of all-in strategic pivot, it validates the fundamental thesis that humanoid robots represent a massive market opportunity. Morgan Stanley projects the global humanoid robot market could reach $38 billion by 2035 and $5 trillion by 2050.
Dan Ives of Wedbush Securities calls Tesla "the best physical AI company in the world" and predicts Tesla could reach a $2 trillion market cap by end of 2026 based primarily on FSD and robotics growth—a 25% stock increase from current levels.
For the humanoid robotics industry as a whole, Tesla's pivot is a legitimizing event comparable to Apple entering the smartphone market. Even if Tesla stumbles on execution, their commitment signals to investors, suppliers, and talent that this market is worth pursuing.
While Tesla restructures for its robot ambitions, China has already established commanding market dominance.
Nearly 90% of all humanoid robots sold globally in 2025 were Chinese. Six of the highest-selling companies in the sector came from China.
Here are the 2025 unit sales according to market research firm Omdia:
That's right: two Chinese companies each outsold Tesla's entire 2025 production target of 5,000 units—a target Tesla failed to meet.
Unitree isn't standing still. The company's CEO Wang Xingxing announced they're targeting 20,000 humanoid robot shipments in 2026—nearly four times their 2025 output. The company is also preparing for a mid-2026 IPO, which would provide additional capital for expansion.
At China's Spring Festival Gala on January 28, 2026, Unitree's robots performed martial arts routines, 3-meter aerial flips, and trampoline somersaults—demonstrating capabilities that put Tesla's awkward walking demos to shame. Their G1 humanoids performed the kung fu sequence without any human intervention at the backend.
For buyers comparing options today, the Tesla Optimus vs Unitree G1 comparison shows just how competitive the pricing landscape has become.
While Unitree targets volume, Figure AI is focusing on enterprise deployments. The BMW manufacturing partnership continues, and Figure's approach of "walking before running"—deploying robots in controlled industrial settings before consumer markets—may prove more prudent than Musk's ambitious consumer-robot vision.
Norwegian company 1X has opened preorders for its NEO humanoid robot, with first customer deliveries planned for 2026. NEO is specifically designed for home use, targeting everyday tasks in unstructured residential settings rather than factory floors. This could give 1X a first-mover advantage in the consumer segment that Musk has promised but not delivered.
China now has over 150 robotics companies actively developing humanoid robots, compared to roughly 20 in the United States. "China is very good at AI, very good at manufacturing, and will definitely be the toughest competition for Tesla," Musk acknowledged at Davos.
Based on Tesla's announcements and historical track record, here's a realistic assessment:
Tesla's Official Timeline:
Reality Check:
Tesla promised 10,000 robots by end of 2025 and likely produced a few hundred. The company has yet to demonstrate an Optimus doing sustained, useful work without teleoperation (human remote control). Multiple supply chain reports throughout 2025 indicated Tesla's Optimus program was "in shambles," with the head of the program departing and production being delayed.
Analyst consensus suggests meaningful commercial production is more likely 2027-2028, with consumer-ready units arriving in late 2028 at earliest.
If you're a consumer interested in purchasing a humanoid robot, Tesla's pivot actually complicates your timeline:
The Good News:
The Bad News:
Our Recommendation:
If you're dead-set on a humanoid robot for home use, watch the 1X NEO closely—they're the most credible consumer play with actual delivery dates. For those willing to wait for Tesla, temper expectations: plan for 2028-2029 for a genuinely useful consumer product, not 2027.
For a comprehensive comparison of all available options, see our best humanoid robots guide and pricing breakdown.
Tesla's decision to end Model S and Model X production represents more than retiring two car models. It's a fundamental reorientation of a company that changed the automotive industry, now betting it can change the robotics industry too.
The Model S proved something that's now easy to take for granted: EVs can work, and ordinary people might actually want one. Now Tesla is attempting to prove something far more uncertain: that humanoid robots can work, and ordinary people (or at least ordinary factories) might actually want them.
Whether this pivot succeeds depends on whether Tesla can:
But even if Tesla stumbles, their commitment has permanently changed the industry's trajectory. When the world's most valuable automaker abandons its flagship vehicles to build robots, it signals to every investor, entrepreneur, and engineer that the humanoid robotics market is no longer science fiction—it's an emerging industry worth betting on.
The most important car of the 21st century is gone. What replaces it will define not just Tesla's future, but potentially the future of work itself.
FF Futurist review with full specs and the truth about its AgiBot A2 origins. Is Faraday Future's full-size humanoid worth $39,990? Expert analysis.
This review covers the full specs, real-world capabilities, pricing, and whether the FF Futurist is worth your money given its origins.
Before diving into specs, you need to understand what you're actually buying.
The FF Futurist is a rebranded AgiBot A2. AgiBot (also known as Zhiyuan Robotics) is one of China's largest humanoid robot manufacturers, having shipped over 5,000 robots by early 2026. When Faraday Future launched its robotics division at the NADA Show in Las Vegas (February 2026), industry observers immediately recognized the hardware.According to Humanoids Daily:
"Despite FF's branding, the hardware appears to be white-labeled versions of the A2 and X2 models developed by Shanghai-based AgiBot."
The AgiBot A2 made headlines in 2025 when it completed a 106km autonomous trek between Suzhou and Shanghai — demonstrating the platform's endurance capabilities. The FF Futurist shares this same hot-swappable battery architecture.
FF's own SEC filings acknowledge a "reliance on a single OEM for robotics products" and "tariff uncertainty for products imported... particularly China."The Ecosystem Skill Package includes additional software capabilities for professional applications.
The FF Futurist emphasizes strength over speed:
This profile suggests the Futurist is designed for tasks requiring force application rather than rapid movement — industrial inspection, object handling, and sustained operation.
Unlike the FF Master, the Futurist includes dexterous hands with tactile sensing as standard:
The sensor suite is more comprehensive than the Master:
FF's "Embodied Intelligence" features include:
The NVIDIA Jetson Orin (200 TOPS) provides more compute than the Master's Orin NX (157 TOPS), enabling more sophisticated on-device AI.
Full disclosure requires acknowledging FF's history:
If you need full human-scale with industrial-grade torque, the FF Futurist is the choice. For budget-conscious or home applications, the FF Master (or direct AgiBot X2) may suffice.
The FF Futurist costs $34,990 base or $39,990 with the optional Ecosystem Skill Package. This positions it as a mid-range professional humanoid — more expensive than the FF Master ($22,990) but significantly cheaper than industrial platforms like the AgiBot A2 direct ($100,000+).
No. The FF Futurist hardware is manufactured by AgiBot in Shanghai, China. It's a white-labeled version of the AgiBot A2. Faraday Future provides the software layer, branding, and U.S. sales/support. FF's SEC filings acknowledge "reliance on a single OEM for robotics products."
The FF Futurist stands 169 cm (5'7" or 66.5 inches) tall and weighs 69 kg. This is full human scale — taller than the FF Master (131 cm) but slightly shorter than the original AgiBot A2 (175 cm). The height difference may indicate a modified configuration.
The FF Futurist is larger (169cm vs 131cm), more powerful (500 Nm vs 120 Nm torque), and includes dexterous hands with tactile sensing as standard. The Futurist is designed for professional/commercial applications, while the Master targets home and education. Price difference: $39,990 vs $22,990.
Faraday Future announced deliveries would begin by end of February 2026. Given FF's history of delayed deliveries in their EV business, buyers should confirm actual availability before committing funds.
Yes. Unlike the FF Master, the FF Futurist includes dexterous hands with tactile sensing as standard equipment. This enables manipulation of objects with feedback sensing. However, the specific payload capacity has not been publicly disclosed by FF.
The FF Futurist is a full-size professional humanoid robot built on proven AgiBot A2 hardware. At $39,990, it offers impressive specs: 500 Nm torque, dexterous hands with tactile sensing, hot-swap batteries, and comprehensive sensors. The "FF Embodied Intelligence" software and U.S.-based support could provide value over direct China imports.
However, transparency requires acknowledging the full picture:*Last updated: February 2026. Specs sourced from FF official announcements and cross-referenced with AgiBot A2 specifications. Robozaps is a humanoid robot marketplace committed to transparent, accurate product information.*
FF Master review with full specs, pricing, and the truth about its AgiBot X2 origins. Is Faraday Future's humanoid robot worth $22,990? Expert 2026 analysis.
This review covers the full specs, real-world capabilities, pricing breakdown, and whether the FF Master is worth your money given its origins.
Before diving into specs, you need to understand what you're actually buying.
The FF Master is a rebranded AgiBot X2, also known as the Lingxi. AgiBot is one of China's largest humanoid robot manufacturers, having shipped over 5,000 robots by early 2026. When Faraday Future launched its robotics division at the NADA Show in Las Vegas (February 2026), industry observers immediately recognized the hardware.According to Humanoids Daily:
"Despite FF's branding, the hardware appears to be white-labeled versions of the A2 and X2 models developed by Shanghai-based AgiBot."FF's own SEC filings acknowledge a "reliance on a single OEM for robotics products" and "tariff uncertainty for products imported... particularly China."
This isn't necessarily bad—it's how many tech products work (see: most Android phones). But transparency matters when you're spending $20,000+.
The Ecosystem Skill Package includes additional software capabilities, though FF hasn't fully detailed what's included vs. base functionality.
The FF Master sits between the budget Unitree G1 and premium options. However, you may be able to source the underlying AgiBot X2 for less through other channels—worth investigating if price is your primary concern.
The FF Master walks at up to 2 m/s (4.5 mph)—faster than average human walking speed. With 30 DOF and 5mm motion precision, it handles:
The AgiBot X2 platform has demonstrated backflips and dynamic acrobatics in controlled settings, though it's unclear if FF enables these capabilities out of the box.
The standard configuration doesn't include dexterous hands—these are an "expandable" option. Without hands, manipulation is limited. This is a key consideration if you need the robot to handle objects.
FF's main value-add is the "FF Embodied Intelligence" software layer:
The NVIDIA Jetson Orin NX (157 TOPS) provides solid on-device AI compute, enabling real-time vision processing and decision-making.
The sensor suite is comprehensive:
Full disclosure requires acknowledging FF's history:
This doesn't mean the FF Master is a bad product—the underlying AgiBot X2 is proven hardware. But buying from FF carries company risk that buying directly from AgiBot or Unitree wouldn't.
FF launched three robots simultaneously:
If you need full human-scale, consider the FF Futurist (or go direct to AgiBot A2).
The FF Master costs $19,990 base or $22,990 with the optional Ecosystem Skill Package. This positions it as a mid-range humanoid robot—cheaper than the Unitree H1 ($90,000) but more expensive than the Unitree G1 ($13,500).
No. The FF Master hardware is manufactured by AgiBot in Shanghai, China. It's a white-labeled version of the AgiBot X2 (Lingxi). Faraday Future provides the software layer, branding, and U.S. sales/support. FF's SEC filings acknowledge "reliance on a single OEM for robotics products."
The FF Master stands 131 cm (4'3" or 51.6 inches) tall and weighs 39 kg. It's a compact, child-sized humanoid—not full human scale. For a taller option, consider the FF Futurist (169 cm) or Unitree H1 (180 cm).
The base FF Master configuration does not include dexterous hands—these are an optional upgrade. Without hands, the robot cannot manipulate objects. If manipulation is important for your use case, factor in the additional cost of hand upgrades.
Faraday Future announced deliveries would begin by end of February 2026. However, given FF's history of delayed deliveries in their EV business, buyers should maintain realistic expectations and confirm actual delivery timelines before committing funds.
The FF Master offers legitimate value as a mid-priced humanoid robot with proven AgiBot X2 hardware. However, buyers should consider: (1) whether the FF branding premium over direct AgiBot purchase is worth it, (2) comfort with FF's company risk, and (3) whether alternatives like the Unitree G1 might better suit their needs at a lower price.
The FF Master is a capable, mid-range humanoid robot built on proven AgiBot X2 hardware. At $22,990, it offers competitive specs: 30 DOF, 2 m/s speed, comprehensive sensors, and solid AI compute. The "FF Embodied Intelligence" software and U.S.-based support could provide value over direct China imports.
However, transparency requires acknowledging the full picture:*Last updated: February 2026. This review reflects information available at publication. Specs sourced from FF official announcements and cross-referenced with AgiBot X2 specifications. Robozaps is a humanoid robot marketplace committed to transparent, accurate product information.*