By Paul Derrington, Jr.
CEO | Chief Engineer | Research Engineer | Editor, D-CSIL
4 min read
Key Takeaways
AGIBOT introduced Worken on October 29 at its Innovation Day in Shanghai, a proposed way to measure robot productivity as Effective Intelligent Work Hours. The formula multiplies active robots by autonomous working hours by an output efficiency ratio, measured against a reference benchmark for a defined task, such as the acceptable work a human completes under established operating standards. AGIBOT, which says its 20,000th robot rolled off the production line in September, wants the industry to look beyond how many robots are built or delivered. Worken is a proposal, not a standard, and AGIBOT says the task-specific benchmarks behind it will be developed from deployment data.
From counting robots to counting work
AGIBOT says its 20,000th robot rolled off the production line in September 2026. At Innovation Day 2026 in Shanghai on October 29, the company said that alongside manufacturing scale and deployment milestones, it wants to measure the productive work robots deliver once deployed. As robots begin working in factories, warehouses, and service settings, AGIBOT says, the practical question changes.
Co-founder, president, and chief technology officer Peng Zhihui put it as a question: “how much productive work are they actually delivering?” Worken is AGIBOT's proposed answer. “We want to move beyond measuring how many robots are built or delivered and begin measuring how effectively they can perform useful work, autonomously and consistently, in the real world,” he said.
Three factors, and each one excludes something
AGIBOT defines Worken as Effective Intelligent Work Hours: the number of active robots, times autonomous working hours, times an output efficiency ratio. Active robots means units performing tasks in real-world environments. Units manufactured or held in inventory do not count.
Autonomous working hours means time spent working on its own, not total powered-on time. The output efficiency ratio compares productive output against a reference benchmark for a defined task, such as the acceptable work a human worker completes under established operating standards. AGIBOT says deployments with similar robot counts can deliver very different productivity depending on how independently the machines operate and how often people must intervene.
The company also ties the metric to lifecycle economics. Hardware depreciates, AGIBOT argues, while software and model updates may let the same machine do more useful work over its operating life. Measuring output alongside total lifecycle cost, it says, could help buyers judge the longer-term value of a deployment.
The hardware and models behind the metric
AGIBOT presented Worken alongside hardware, model, and software releases that it says could make such measurement more practical. AHA3, AGIBOT's third-generation humanoid hardware architecture, integrates the hardware foundations for locomotion, interaction, and task execution into one platform, with Wi-Fi 6 and 5G connectivity, a joint liquid-cooling system, IP67-rated joints, and a dexterous-hand family including OmniHand 3 Lite, OmniHand 3, and OmniHand 3 Ultra plus the OmniPicker 3 tactile gripper.
On models, AGIBOT debuted GO-3 Preview, an embodied foundation model built around a unified reasoning, prediction, and action framework called WRAM, the World Reasoning Action Model, and showcased WITA-Omni, an omni-modal interaction model for full-duplex conversation with coordinated speech and movement and on-device deployment. AGIBOT says GO-3 Preview ranked first in task success rate and process score on RoboDojo, achieved the highest overall score on RoboTwin, and ranked first overall on RoboColiseum, and that WITA-Omni ranks first on embodied interaction benchmarks including Daily-Omni. Those results were presented at AGIBOT's own event. RoboDojo's public leaderboard, which a non-profit academic group maintains, lists an entry named GO-3 Nano Preview in first place. RoboColiseum is built on AGIBOT's own Genie Sim Benchmark, so it is not independent of the company. The RoboTwin and Daily-Omni results could not be confirmed on the public pages checked for this story.
The company also formally launched G2 Max, a heavy-duty industrial robot with 21 degrees of freedom, dual-arm handling capacity of up to 40 kg, and an adjustable working height from 0.1 to 2.5 meters. AGIBOT pointed to deployments of its wider G2 family at Longcheer tablet production, semiconductor packaging and testing, CNC machine tending, and logistics, and to service deployments of its robots at Guangzhou Metro and the Chimelong tourism and entertainment complex.
What Worken is not yet
Worken is a proposed framework, and AGIBOT describes it as evolving. AGIBOT says it will develop the task-specific benchmarks Worken depends on from real-world deployments and operational data, and that it hopes the measure can eventually serve as a widely applicable reference. Worken also appears in the 2026 Embodied Intelligence Industry Innovation Report, compiled by China Economic Information Service and the Shanghai Industrial Innovation Center of the China Academy of Information and Communications Technology, which was released at the event.
AGIBOT also released Link-U OS 1.0, which it positions as an operating system purpose-built for embodied intelligence, now being trialled by selected ecosystem partners with plans to become fully open source by the end of 2026. The company donated the project to the OpenAtom Foundation, became a Platinum Level sponsor, and says its ecosystem spans 255 universities, 511 enterprises, and more than 80,000 developers. Open questions are whether those task benchmarks are published, and whether organizations outside AGIBOT adopt the measure.