Unitree led on the strictest comparison: about 18,000 completed upright bipedal humanoids versus AgiBot’s reported 7,000 pure bipeds. The figures suggest a two tier market: Unitree and AgiBot operate at the ten thousand unit scale, while UBTECH and Galbot remain closer to the thousand unit range.
Research answer

Create a landscape editorial hero image for this Studio Global article: How, as of July 2026, does Unitree’s cumulative production of 18,000 complete upright bipedal humanoid robots—excluding wheeled robots, quad. Article summary: On the stated definitions, Unitree is the clear leader in completed upright bipedal humanoids: 18,000 units is about 2.6 times AgiBot’s reported 7,000 pure bipeds, six times UBTECH’s roughly 3,000 all-series total, and 1. Topic tags: general, news, general web, user generated. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts w
Unitree’s reported production milestone changes the comparison among China’s leading humanoid-robot makers. As of July 2026, it had produced approximately 18,000 complete bipedal humanoids, excluding wheeled platforms, quadrupeds, and standalone robotic arms. That gives it a clear lead over AgiBot’s reported 7,000 pure bipeds—but not necessarily over AgiBot on every measure of market performance. 8
18
The headline numbers use different definitions:
On a like-for-like bipedal basis, Unitree’s 18,000 units are roughly 2.6 times AgiBot’s reported 7,000. They are also about six times the approximately 3,000-unit UBTECH figure and 18 times Galbot’s approximately 1,000.
That does not make wheeled robots irrelevant. A wheeled dual-arm system can be more stable, less expensive, and easier to deploy for some factory or logistics tasks. It simply should not be treated as equivalent to a robot designed to walk upright through human-oriented environments.
The available figures point to a clear production gap.
Both companies have reported cumulative output in the tens of thousands, although their product mixes differ. AgiBot reached a reported 15,000 cumulative general-purpose humanoid robots in June 2026, with the milestone unit identified as a G2 industrial robot. 38
40 Unitree’s later 18,000 figure is narrower because it counts bipedal humanoids only.
18
48
Shipment estimates reinforce the importance of these two companies. Counterpoint reported that Galbot exceeded 1,100 shipments in the first half of 2026 and that UBTECH reached the thousand-unit level, while separate estimates placed AgiBot and Unitree far ahead of the rest. 2
5
UBTECH has meaningful industrial momentum, but production capacity should not be confused with cumulative output. The company said Walker S2 entered mass production and delivery in 2025 and disclosed annualized capacity of more than 6,000 full-size embodied-intelligence humanoids at the end of that year. 4
Galbot’s reported H1 shipments above 1,100 units show commercial traction, particularly in industrial and logistics applications. Its S1, launched in January 2026, uses a wheeled dual-arm design aimed at heavy-load industrial and logistics scenarios. 2
The gap between these companies and the two leaders is therefore real in reported volume, but it is not necessarily a simple ranking of technical quality or business potential. Different product strategies serve different customer requirements.
Unitree’s production trajectory is the most striking part of the milestone. The company reportedly produced about 6,500 units in 2025, passed 11,000 G1 units by June 2026, and later reported approximately 18,000 cumulative bipedal humanoids. 46
48
If those figures are measured on compatible bases, the increase from more than 11,000 to 18,000 represents roughly 7,000 additional bipeds in a short period. That would indicate a rapid manufacturing ramp rather than a business still limited to one-off prototypes.
The important limitation is what the number does not show. Production, shipments, customer orders, deployed fleets, and robots actively performing useful work are separate metrics. A factory milestone does not by itself establish utilization, revenue, autonomy quality, safety, reliability, or unit economics.
The evidence supports a manufacturing strategy built around product standardization and control of key hardware. Unitree describes itself as focused on the research, production, and sale of legged and humanoid robots, manipulators, and related systems. 30 Industry reporting also describes its humanoid platforms as using standardized joint modules and modular compute architecture, with the company developing core actuation hardware in-house.
23
24
That approach can provide several potential advantages:
These are reasonable explanations for Unitree’s ability to produce at scale, but the sources do not independently quantify how much each factor contributed to the 18,000-unit result. It would therefore be too strong to attribute the milestone to a single advantage such as servo motors, drivers, reducers, or factory layout.
Unitree’s product segmentation may also help it reach different buyers. The G1 is positioned as a compact platform for research, robotics development, and automation, while the H1 targets larger-scale locomotion and higher-performance applications. 17
19 A lower-cost, developer-oriented product can potentially support higher unit volume, while a full-size platform addresses more demanding mobility and industrial requirements.
AgiBot’s 15,000-unit milestone demonstrates substantial manufacturing speed. The company said it moved from 10,000 to 15,000 units in roughly 90 days, and the milestone robot was a G2 industrial task robot. 38
40
Its broader product mix is strategically useful because it covers more than one form factor and operating environment. Reporting describes AgiBot’s portfolio as spanning full-size bipedal, compact bipedal, and wheeled systems, while Counterpoint’s shipment estimates placed it ahead of Unitree in H1 2026 shipments. 5
44
That creates two different answers to the question of who is leading:
The result is not a contradiction. It reflects the difference between a narrow production count and a broader shipment measure.
It would be misleading to describe UBTECH as incapable of mass production. Its Walker S2 entered mass production and delivery, and the company reported annualized capacity above 6,000 full-size humanoids at the end of 2025. 4
What the evidence does support is a distinction between capacity, output, orders, and deliveries. UBTECH’s reported capacity and order book show that it is preparing to scale, but those figures cannot be used as proof that it has already produced or deployed the same volume as Unitree or AgiBot. Reuters separately reported 13,361 U1-series orders in July 2026, which are orders rather than completed deliveries. 1
The supplied evidence does not establish that UBTECH’s relative position is specifically the result of an early focus on small consumer robots, outsourced core components, or fragmented resources. Those explanations should remain hypotheses rather than stated conclusions. Its industrial transition is documented; the precise reasons for the production gap are less certain.
The strongest conclusion is that China’s humanoid-robot sector has moved beyond a purely prototype-driven phase. Companies are reporting multi-thousand-unit production, commercial orders, and factory deployments, while an industry estimate cited by Gasgoo said annual production of complete humanoid robots could exceed 100,000 units in 2026. 41
But the evidence does not yet prove a durable Unitree–AgiBot duopoly. The two companies clearly form the leading scale group in the figures reviewed, yet AgiBot leads some current shipment estimates while Unitree leads the narrower pure-biped cumulative-production comparison. 2
5
18
Nor does the data prove that competitors face a fixed two- or three-year catch-up period. Closing the gap will depend on capital, component supply, factory yield, product standardization, customer demand, service infrastructure, and the ability to turn production into reliable real-world work.
The next phase of competition will therefore be measured by more than robots coming off an assembly line. The decisive questions are whether these machines can operate safely, complete useful tasks consistently, remain serviceable at scale, and generate economics that justify wider deployment.
Studio Global AI
This page includes a source-backed answer you can continue inside Studio Global.
Unitree led on the strictest comparison: about 18,000 completed upright bipedal humanoids versus AgiBot’s reported 7,000 pure bipeds.
Unitree led on the strictest comparison: about 18,000 completed upright bipedal humanoids versus AgiBot’s reported 7,000 pure bipeds. The figures suggest a two tier market: Unitree and AgiBot operate at the ten thousand unit scale, while UBTECH and Galbot remain closer to the thousand unit range.
The milestone proves manufacturing scale, not necessarily deployment success, autonomy, reliability, revenue, or favorable robot economics.