The AgiBot X2 Series (marketed under the Lingxi brand name in China) is a family of compact, full-body bipedal humanoid robots developed by AGIBOT Innovation Shanghai Technology Co., Ltd., the world's highest-volume humanoid robot manufacturer by units shipped in 2025. Officially unveiled on March 11, 2025, by AgiBot co-founder and CTO Peng Zhihui, the Lingxi X2 was introduced as a generalist humanoid robot designed to demonstrate the "perfect fusion of artificial intelligence and humanoid robotics technology" across motion intelligence, interaction intelligence, and task intelligence simultaneously in a single compact platform.

AgiBot X2 Series

5 Items

AgiBot X2 Series USA: The Lingxi Compact Humanoid Family Explained

Standing 1.3 meters tall and weighing 33.8 kilograms, the X2 is intentionally sized between the smallest companion robots and full-size humanoids like AgiBot's own A2 Ultra. This form factor was chosen deliberately: it allows the robot to navigate stores, hospitals, classrooms, and home environments where full-size humanoids at 1.7 meters or taller would be physically challenging to operate or too visually imposing for comfortable human interaction.

The X2 Series currently encompasses four variants: the standard X2 (entertainment and commercial performance platform), the X2 Ultra (research and autonomous deployment flagship), the X2-N (hybrid bipedal-wheeled locomotion platform), and the X2-W (wheeled variant in development). Together, the series forms AgiBot's compact humanoid product family, positioned below the A2 Series in both price and size while providing a broader range of interaction, education, and research capabilities than most robots in its size class.

AgiBot and the X Series

AgiBot was founded in February 2023 in Shanghai by Deng Taihua (CEO) and Peng Zhihui (CTO), both formerly of Huawei's elite Genius Youth Program. The company grew to ship over 5,100 humanoid robots in 2025, making it the world's number-one humanoid manufacturer by volume according to analyst firm Omdia. The company's 10,000th robot rolled off the production line in March 2026.

The X Series, known internally as the Lingxi line, was designed in parallel with the company's flagship A Series (full-size service humanoids) and G Series (industrial wheeled humanoids) to address a distinct use case: compact, agile humanoid platforms that can be deployed in education, research, entertainment, and commercial interaction settings where the cost, size, and operational requirements of larger platforms are prohibitive.

The X2 represents the second generation within the Lingxi line, following the earlier X1 and X1-W research platforms. The X2 was developed by AgiBot's X-Lab, the company's dedicated compact humanoid research division. Peng Zhihui personally introduced the X2 through a widely shared video demonstration in March 2025, which showed the robot walking, running, dancing, riding a bicycle, balancing on a hoverboard, and reading medication instructions aloud, establishing the platform's public profile.


X2 Series Variants

X2 Standard: The Commercial Entertainment Platform

The X2 Standard is described by AgiBot on its official store as a "half-size humanoid robot engineered for the entertainment and commercial performance sector." It supports a wide variety of full-body motion transitions and provides a rich set of preset motion skills. Its flexible material shell and compact, lightweight construction prioritize safety and reliability in environments shared with humans.

The X2 Standard features 25 to 28 degrees of freedom, including one degree of freedom in the head and three in the waist, enabling versatile and agile motion with vivid, lifelike expressions. It supports multimodal interaction including visual, voice, tactile, and facial expression capabilities, fusing user gestures, environmental data, and contextual information into advanced intelligent dialogue. The platform is driven by dual Rockchip RK3588 processors as its primary compute units. The RK3588 is an octa-core ARM system-on-chip with an integrated 6 TOPS neural processing unit, handling the operating system, interaction logic, voice processing, facial expression control, screen rendering, and communication protocols.

The X2 Standard is positioned for commercial deployments where the primary requirement is expressive, engaging physical performance and natural conversational interaction, rather than autonomous navigation or complex manipulation.

X2 Ultra: The Research and Autonomous Deployment Flagship

The X2 Ultra is the flagship variant within the X2 family and is described by AgiBot as the platform for "power users, researchers, and high-stakes service environments." It shares the same kinematic skeleton and industrial design as the X2 Standard but adds a significant hardware upgrade in the form of an NVIDIA Jetson Orin NX AI accelerator delivering 157 TOPS of compute power alongside the dual RK3588 base compute. This triple-processor architecture provides sufficient onboard AI compute to run deep learning models for spatial mapping, obstacle avoidance, object recognition, and manipulation planning simultaneously without cloud dependency.

Beyond compute, the X2 Ultra adds hardware not present in the standard model: a 3D LiDAR sensor for 360-degree spatial awareness and SLAM-based autonomous navigation, an RGB-D depth camera for precise close-range object detection and manipulation, front binocular stereo vision cameras, a rear monocular RGB camera, 4G/5G cellular connectivity for remote management and OTA updates in the field, and an optional autonomous charging dock that allows the robot to return to its station and recharge independently when battery power falls below a threshold.

The X2 Ultra also expands the robot's degrees of freedom to 30 to 31, adding a neck degree of freedom and improving arm articulation compared to the standard variant. It is compatible with AgiBot's OmniHand dexterous manipulation system and OmniPicker end-effectors, which are sold separately and extend the platform's utility for manipulation research and task execution applications. The X2 Ultra runs Lingqu OS, described by AgiBot as the world's first embodied intelligent operating system, released in July 2025. Lingqu OS provides a layered, open-source robotics software stack built on top of AimRT, AgiBot's proprietary real-time middleware that operates as an alternative to ROS 2 DDS implementations with up to 30 percent lower latency. The system supports OTA updates, enabling operators to push new capabilities to deployed units remotely.

X2-N: The Hybrid Bipedal-Wheeled Platform

The X2-N is a technically distinctive variant that adds wheel-foot locomotion capability to the X2's bipedal body. It features integrated foot wheels that allow the robot to transition seamlessly between standard bipedal walking and wheeled rolling, enabling it to traverse flat surfaces efficiently on wheels while retaining the ability to walk bipedally on stairs, slopes, and complex terrain where wheels are impractical.

Uniquely, the X2-N is reported to operate its proprioceptive navigation without relying on cameras or external sensors for locomotion decisions, instead using joint torque feedback, pressure sensors, and internal gyroscopes for real-time terrain adaptation. This approach produces a system that can carry cargo, adapt its gait autonomously, and navigate dynamic, unpredictable environments including logistics facilities and search-and-rescue settings.

The X2-N was the primary CES 2026 showcase robot for the X Series, performing Tai Chi routines at AgiBot's booth in the Las Vegas Convention Center's North Hall and drawing strong audience interest. Reports identify the X2-N as leading shipment volumes among the X2 variants, suggesting it has become the most commercially active platform within the series.

X2-W: The Wheeled Compact Variant

The X2-W is a wheeled variant of the X2 platform, replacing bipedal legs with a wheeled mobile base for environments requiring sustained operational stability on flat floors. Detailed specifications for the X2-W were not publicly disclosed at the time of the X2 Series' release, but it shares the interaction AI stack of the rest of the X2 family and is designed for commercial and service deployment scenarios.


Technology and Specifications

Physical Specifications

The X2 standard and Ultra variants share a common body: 1.3 meters tall, weighing 33.8 kilograms. The battery provides approximately two hours of runtime at 0.5 meters per second walking speed, with a battery capacity of approximately 500 Wh. The X2 Ultra supports an interchangeable battery for operational continuity. Maximum running speed for the platform is 1.5 meters per second.

Degrees of freedom are 25 to 28 for the X2 Standard and 30 to 31 for the X2 Ultra, with the additional joints in the neck and arms. The peak joint torque of 120 Nm supports the dynamic motion capabilities including running, dance routines, stair climbing, and bicycle riding. The flexible exterior uses thermoplastic polyurethane foam wrapping for human-safe physical interaction.

The Xyber Component System

The X2 is built on a proprietary modular electronics architecture that AgiBot calls the Xyber system. This architecture consists of four interdependent subsystems that collectively handle the robot's real-time control from the perception layer to the actuator level.

The Xyber-Edge is the cerebellum controller, responsible for low-level motion regulation, balance maintenance, and the fine-tuned joint coordination required to execute dynamic sequences like bicycle riding or hoverboard balancing. Operating analogously to the human cerebellum, it processes proprioceptive feedback from joint encoders and inertial sensors at high frequency to maintain stability across all of the robot's 28 to 31 degrees of freedom in real time.

The Xyber-DCU is the domain controller, handling higher-level decision-making and motion planning. When the robot receives a task instruction, the Xyber-DCU decomposes it into a sequence of motion primitives and coordinates execution across the robot's limbs, managing the workflow from intention to physical action.

The Xyber-BMS is the intelligent battery management system, monitoring cell-level charge status, temperature, and discharge rates across the battery pack. It manages energy allocation during demanding motion sequences to prevent sudden power drops during high-torque activities like running or stair climbing.

The Powerflow system is AgiBot's proprietary core joint module set, the servo actuators that execute the joint movements commanded by the Xyber-Edge and Xyber-DCU. These are custom-developed components rather than off-the-shelf servo units, which allows AgiBot to tune torque delivery, compliance, and backdrivability characteristics specifically for the dynamic and dexterous motion requirements of the X2's application profile.

Silicon Photonic Visual Language Model

One of the X2's most technically distinctive elements is its Visual Language Model (VLM), which uses silicon photonic technology for visual processing. Unlike traditional GPU-based computer vision systems that use electronic circuits to process image data, silicon photonic systems use optical circuits that transmit data as photons of light rather than electrons. This approach enables significantly faster processing at lower power consumption than electronic equivalents.

For the X2, the silicon photonic VLM enables sub-millisecond visual processing that is faster than comparable GPU-based systems. In practice, this means the robot can perceive, interpret, and respond to visual information from its environment in near-real time, a capability that was demonstrated in the official X2 launch video when the robot read printed medication instructions aloud, requiring it to identify the text, parse its meaning, and speak the content fluently in a single rapid sequence.

WorkGPT and Interaction Intelligence

WorkGPT, AgiBot's proprietary large language model, handles natural language understanding and task planning for the X2. When a user speaks to the X2, WorkGPT processes the speech input, determines the intent, plans the appropriate response or action sequence, and coordinates output across the robot's voice synthesis, facial expression system, and physical motion system simultaneously. This unified pipeline produces natural, coherent interaction that responds appropriately to conversational context rather than executing isolated command-response pairs.

AgiBot describes the X2 as the "first truly interactive dynamic robot," capable of responding to human actions within milliseconds. The combination of silicon photonic visual processing, WorkGPT language understanding, and the Xyber-Edge's real-time balance control enables the robot to maintain dynamic physical activities such as bicycle riding while simultaneously processing verbal interaction from a nearby person, a capability that most robots achieve only in one domain at a time.


Applications and Use Cases

Education and University Research

The X2 Ultra's open development environment, secondary development support, dual LAN ports, expanded USB connectivity, OTA update capability, and Lingqu OS open-source software stack make it a practical research platform for university engineering programs and robotics research groups. The GO-1 foundation model and AgiBot World dataset are compatible with the X2 Ultra, enabling research groups to deploy and evaluate embodied AI models on physical hardware without writing robot-specific control code from scratch.

The robot's dynamic capabilities, including stair climbing, bicycle riding, hoverboard balancing, and a Webster Flip, make it a compelling hardware platform for locomotion research, where the breadth of achievable motion is as important as the quality of any single skill.

Healthcare and Elder Care

The X2's compact 1.3-meter height allows it to operate in hospital corridors, patient rooms, and care facility spaces where full-size humanoids would be physically impractical or socially imposing. The demonstrated capability to read medication instructions aloud, a task requiring simultaneous vision, language processing, and speech synthesis, has direct clinical relevance for visually impaired patients or individuals with reading difficulties who need assistance interpreting prescription labels or care documentation.

In elder care contexts, the X2's natural conversational interaction, expressive facial and body behavior, and compact non-threatening form factor support engagement, routine reminders, and companionship functions. The X2 Ultra's autonomous navigation and autonomous charging dock enable sustained unattended operation in a care facility without requiring staff to manually position or recharge the robot.

Retail and Exhibition

The X2 is positioned by AgiBot as suitable for in-store reception, exhibition guidance, cultural and entertainment tourism, and brand ambassadorship. At 1.3 meters, the robot operates at a height that facilitates natural eye contact with seated visitors, children, and people at reception desks, while remaining compact enough to navigate between retail display fixtures and exhibition stands.

The X2 Standard's rich preset motion library and entertainment-focused design make it particularly effective for creating memorable in-store experiences, where physically dynamic behavior, including dance, expressive gesture, and coordinated movement, draws visitor attention more effectively than a stationary display.

Performance Art and Live Events

The X2's dance capabilities, fluid motion transitions, and synchronized movement with voice output make it a natural fit for live performance contexts including corporate events, product launches, and cultural shows. AgiBot Night in February 2026 demonstrated the broader AgiBot fleet in live performance, with the X2 family contributing to the performance segments.

Home Assistance

Notwithstanding the current state of commercial maturity, AgiBot positions the X2 Ultra as a candidate for home assistance applications including light household tasks, companion care for older adults, and home security monitoring. The robot's compact size, autonomous charging capability, and natural interaction system support these scenarios more practically than full-size humanoids that would require home modifications to operate.


Advantages and Benefits

Silicon photonic VLM for sub-millisecond visual processing: The use of optical rather than electronic circuits for the Visual Language Model provides visual processing speeds that exceed traditional GPU-based alternatives, enabling truly real-time scene understanding and responsive interaction.

Compact form factor for real-world deployment: The 1.3-meter, 33.8-kilogram body opens environments to robotic deployment that full-size humanoids cannot practically access, including smaller retail spaces, clinical environments, classrooms, and household settings.

Unique dynamic capabilities: Bicycle riding, hoverboard balancing, scooter operation, Webster Flip, and stair climbing in a sub-35-kilogram platform provide a breadth of dynamic motion that no comparable product in the compact humanoid category has matched.

World's first embodied OS (Lingqu OS): The X2 Ultra's Lingqu OS provides a unified open software environment with up to 30 percent lower latency than ROS 2, OTA update support, and compatibility with AgiBot's full model and development toolkit, making it the most software-capable compact humanoid on the market.

Open research ecosystem: GO-1 model compatibility, AgiBot World dataset access, secondary development support, and Genie Sim 3.0 simulation tools collectively provide X2 Ultra buyers with the most complete open embodied AI research infrastructure available on a commercial compact humanoid platform.

Accessible entry pricing: Starting from approximately $20,000 to $22,000 USD for the standard X2 configuration, the series provides a substantially lower barrier to entry than full-size enterprise humanoids while delivering a significantly richer capability set than toy and educational robots in the $1,000 to $5,000 range.


Comparison with Related Platforms

AgiBot X2 vs. Unitree G1

Unitree's G1 ($21,600) is the most direct competitor to the AgiBot X2 in the compact agile humanoid market. The G1 has strong developer community support, established US distribution, and documented martial arts demonstration capabilities, including synchronized performances at China's Spring Festival Gala. The X2's silicon photonic VLM, Lingqu OS, and bicycle-riding capability distinguish it on the interaction intelligence and dynamic motion fronts. The G1's broader US distribution infrastructure currently provides easier procurement for most US buyers. For those prioritizing research-grade autonomous navigation and manipulation, the X2 Ultra's NVIDIA Jetson Orin NX compute and LiDAR package offer a higher sensor and compute baseline than the standard G1 configuration.

AgiBot X2 vs. Unitree H1

Unitree's H1 ($90,000) is a full-size bipedal humanoid targeting athletic performance at a significantly higher price point. The H1's walking speed and payload capacity exceed the X2's, but it operates in a different deployment category. For buyers in education, healthcare, retail, and entertainment who need a compact, affordable, and socially appropriate humanoid, the X2 Series provides capabilities the H1 does not emphasize (interaction intelligence, compact form factor, VLM-driven conversation) at a fraction of the H1's cost.

AgiBot X2 vs. AgiBot A2 Lite

Within AgiBot's own lineup, the X2 and A2 Lite occupy adjacent market positions. The A2 Lite ($100,000 and above) is a full-size bipedal platform with 23 DOF optimized for service reception and corporate deployment, holding FCC certification for US public spaces. The X2 Ultra at around $20,000 EUR is substantially more affordable, more compact, and better suited for education, research, and entertainment. The A2 Lite's taller stature and FCC certification make it the better choice for corporate lobby reception and regulated public venues. The X2 Ultra's research ecosystem, silicon photonic VLM, and compact size make it the better choice for university labs, healthcare corridors, retail floors, and development projects.


Summary

The AgiBot X2 Series (Lingxi) is the most technically ambitious compact humanoid robot family commercially available to US buyers in 2026. Its silicon photonic Visual Language Model for sub-millisecond visual perception, Xyber intelligent component architecture, 28 to 31 degrees of freedom, bicycle-riding and hoverboard-balancing dynamic capabilities, and the X2 Ultra's Lingqu OS with NVIDIA Jetson Orin NX compute at 157 TOPS collectively represent a capability profile that no directly competing compact humanoid platform had matched at the time of the series' launch. The X2 Standard's entertainment-focused design and starting price around $22,000 make it accessible to retail, education, and healthcare organizations seeking a compact interactive humanoid. The X2 Ultra's open research ecosystem, autonomous navigation, OmniHand compatibility, and full AgiBot software stack support make it the leading candidate for US university research programs and technology developers building next-generation embodied AI applications. Backed by the world's number-one humanoid robot manufacturer by volume, the AgiBot X2 Series is a credible, production-scale platform for anyone seeking to deploy or research compact humanoid robots in the United States today.

Questions

Your Question:

What is the AgiBot X2 Series?

The AgiBot X2 Series, known as the Lingxi Series in China, is a family of compact bipedal humanoid robots developed by AGIBOT Innovation Shanghai Technology Co., Ltd. and officially launched on March 11, 2025. The series includes the X2 Standard (entertainment and commercial interaction platform), the X2 Ultra (research and autonomous deployment flagship), the X2-N (hybrid bipedal-wheeled locomotion platform), and the X2-W (wheeled variant). All variants stand approximately 1.3 meters tall and weigh 33.8 kilograms. The X2 is built on the Xyber component architecture and uses a silicon photonic Visual Language Model for sub-millisecond visual processing, combined with WorkGPT for natural language interaction.

How does the AgiBot X2's silicon photonic VLM work?

The X2's Visual Language Model uses silicon photonic circuits, which transmit and process data as photons of light rather than as electrical signals in conventional electronic circuits. Because light travels faster than electrical signals and optical circuits generate less heat, silicon photonic systems can perform the same computational operations in substantially shorter time with lower power consumption. For the X2, this enables visual data from the robot's cameras to be processed in under a millisecond, allowing the robot to recognize objects, read text, and interpret its visual environment in near-real time. This speed underpins the demonstrated capability of reading medication labels aloud and responding to visual cues mid-interaction without perceptible delay.

Why is the AgiBot X2 important for education and research?

The X2 Ultra is significant for university and research applications because it combines a physically capable and dynamically challenging robotic body with a fully open software and AI ecosystem. The combination of Lingqu OS, AimRT middleware with 30 percent lower latency than ROS 2, NVIDIA Jetson Orin NX at 157 TOPS, GO-1 foundation model compatibility, AgiBot World dataset access, and Genie Sim 3.0 simulation tools provides a more complete research infrastructure on a single commercially available platform than any directly comparable compact humanoid. The X2's breadth of dynamic capabilities, from stair climbing to bicycle riding, also provides a challenging and diverse set of locomotion problems for motion control and reinforcement learning research.

What are the benefits of the AgiBot X2 for healthcare and elder care?

The X2 offers several specific benefits in healthcare and elder care settings. Its 1.3-meter height allows operation in hospital corridors, patient rooms, and care facility spaces that full-size humanoids cannot practically navigate. The demonstrated ability to read medication instructions aloud is directly relevant for visually impaired patients or those with reading difficulties. The X2 Ultra's autonomous navigation and automatic charging dock enable sustained unattended deployment without staff intervention. The robot's natural conversational AI, expressive physical behavior, and compact non-threatening appearance support engagement, routine reminders, and companionship functions. And the flexible exterior material reduces the injury risk from incidental contact in shared spaces with elderly or medically vulnerable individuals.

What is Lingqu OS and why does it matter?

Lingqu OS is AgiBot's proprietary embodied intelligent operating system, released in July 2025 and described by AgiBot as the world's first operating system specifically designed for embodied AI robots. It is built on top of AimRT, AgiBot's C++20 real-time middleware, and provides a layered open-source software stack that achieves up to 30 percent lower latency than ROS 2 DDS implementations. Lingqu OS supports OTA updates, enabling operators to deploy new robot capabilities remotely without physical access to the hardware. For US developers and research teams building applications on the X2 Ultra, Lingqu OS provides a more efficient, better-supported, and lower-latency software environment than adapting standard robotics middleware frameworks, and its open-source layer enables the kind of customization that proprietary closed-stack systems cannot support.