LimX Dynamics

LimX Dynamics of Shenzhen is one of the younger Chinese players that has already gathered serious capital: USD 400 million in total funding at a USD 2.2 billion valuation, and the company is preparing a stock market debut in Hong Kong (HKEX). The founding year and headcount are not given in available sources, so there is no chronicle from day one. The product dates, however, are clear. In July 2025 LimX presented Oli, a full-size research humanoid, and in August 2025 it put it on sale from USD 21,800. The CL-3 designation is an internal working name for the same product, not a separate model. The company formally introduced its own COSA system (Cognitive OS of Agents) only in January 2026, half a year after the hardware launch; at launch the robot relied on general whole-body control algorithms and an open SDK. The basics of the field are explained in “How Does a Humanoid Robot Work”. In dealers' listings both designations may appear side by side, so when comparing prices check which variant (Lite, EDU or Super) is meant.

The portfolio is narrow but coherent: platforms for laboratories. The LimX Oli is a 165 cm humanoid weighing up to 55 kg, with 31 active degrees of freedom (up to 43 with two manipulation hands), a 3 kg payload per arm and about 1.5 hours of runtime on a 558 Wh battery. The LimX Oli entry describes the same product under its working name, with a starting price of USD 21,800 (about PLN 84,500 (≈ EUR 19,360)) for the Lite variant. Equipment includes an in-house IMU, Intel RealSense D435i cameras (Intel RealSense) in the head and chest, and, from the EDU variant, an NVIDIA Jetson Orin computer, while the open Python SDK works with Isaac Sim, MuJoCo and Gazebo. The LimX Dynamics TRON 1 is the cheaper entry: 20 kg weight, 10 kg payload, a 211 Wh battery for around 2 hours, and an Intel Core i3 computer; the price is not disclosed. A related category: “Educational robots”. Note that the onboard computer in the TRON 1 is a general-purpose, laptop-class chip, not a dedicated AI platform. For a lab, that makes it easier to run your own code.

By capitalization LimX is among the top Chinese humanoid startups, though still clearly below Unitree Robotics or Figure AI. Its main argument is price: USD 21,800 for a full-size research humanoid is one of the lowest in this class. The Unitree G1 starts at about USD 16,000, so it is cheaper but smaller; Oli has the full height of an adult and a claimed larger number of hand degrees of freedom. Among Chinese rivals targeting universities is also Fourier Intelligence, whose humanoids are pricier or have unconfirmed prices. Oli's weak points are its short battery life and modest 3 kg payload, which limit manipulation tasks. We have no named customers. For a comparison of price classes, see “Unitree H1, G1, R1 and H2”. The planned Hong Kong listing adds credibility but also pressure on sales results. The open SDK is a further argument for laboratories.

There is no official availability in Poland, and no distributor has been confirmed in Europe. All three models are listed as unavailable in the country. For Polish universities and robotics labs, though, Oli is potentially one of the cheapest entry points to a humanoid platform in this group. A purchase would require ordering directly from the manufacturer, with import, customs duty and servicing left to the buyer. Those looking for a path into teaching and research will find it in “How to Start Learning Robotics in Poland” and laboratories. A university with a research budget could therefore consider ordering a unit, provided it can arrange transport and technical support outside an official channel.

LimX Dynamics is a well-capitalized maker of research platforms with an honestly stated price for Oli. For universities it is an interesting candidate, provided import and support can be arranged. For a company looking for a robot that actually does work, it is too early: a 3 kg payload and 1.5 hours of battery are laboratory-grade figures. An added plus is the open Python SDK, which makes integration with popular simulators easier. Order only after the variants and delivery are clarified.