AGIBOT Proposes Worken, a Metric for Productive Robot Work
AGIBOT has introduced Worken, a proposed framework for measuring how much productive work robots perform in real-world settings, at its Innovation Day 2026 in Shanghai on October 29. The company says the metric is meant to shift attention from the number of robots built or deployed to the useful work they deliver autonomously and consistently.
- AGIBOT proposed Worken, a metric for productive robot work, at Innovation Day 2026 in Shanghai on October 29.
- Worken = Number of Active Robots × Autonomous Working Hours × Output Efficiency Ratio.
- AGIBOT unveiled the AHA3 hardware platform, with liquid cooling, IP67-rated joints and the OmniHand 3 family.
- GO-3 Preview, built on WRAM, ranked first on RoboDojo task success rate and process score, according to results presented at the event.
- The company also launched G2 Max and Link-U OS 1.0 and showed the WITA-Omni interaction model.
What Worken measures
The name combines "Work" and "Token". AGIBOT describes Worken as "Effective Intelligent Work Hours" and defines it with a formula: Worken = Number of Active Robots × Autonomous Working Hours × Output Efficiency Ratio.
- Number of Active Robots: counts robots actively performing tasks in real-world environments, not units that were manufactured or sit in inventory.
- Autonomous Working Hours: measures time spent working autonomously, rather than total powered-on time.
- Output Efficiency Ratio: assesses productive output against a reference benchmark for a defined task, such as the amount of acceptable work completed by a human worker under established operating standards.
According to AGIBOT, deployments with similar numbers of robots can deliver different levels of productivity, depending on how independently the machines operate, how often human intervention is needed and how much acceptable work they complete.
Peng Zhihui, Co-founder, President and CTO of AGIBOT, framed the idea as a practical question for the industry. "Worken is our attempt to develop a new way of answering that question," he said, referring to how much productive work robots are actually delivering.
Economics and next steps for the framework
AGIBOT says Worken could apply to areas such as industrial handling, logistics, inspection and commercial services. The company also links it to the economics of robotics: while hardware depreciates over time, improvements in software and AI models may let the same machine do more useful work during its operating life. Measuring that output alongside total lifecycle costs could help businesses judge the longer-term value of robot deployments, according to the company.
AGIBOT calls Worken an evolving framework that will be refined through real-world deployments and operational data. It plans to develop task-specific benchmarks and hopes Worken can eventually become a widely applicable reference for measuring the productive value of physical AI.
The event also saw the release of 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 (CAICT). The report includes Worken and AGIBOT's deployment examples in its industry analysis.
AHA3 hardware platform and motion intelligence
On the hardware side, AGIBOT presented AHA3 (AGIBOT Humanoid Architecture), its third-generation humanoid hardware architecture and first commercial "Three Intelligences in One" platform. It integrates the hardware foundations for locomotion, interaction and task execution intelligence into one architecture. The platform builds on motor-drive technology, integrated computing and high-speed connectivity, including Wi-Fi 6 and 5G, and is designed to reduce integration complexity and improve system responsiveness.
For sustained industrial use, AGIBOT introduced a joint liquid-cooling system meant to improve thermal performance during continuous operation, plus specialized joints with IP67-rated protection. The new dexterous-hand family includes OmniHand 3 Lite, OmniHand 3 and OmniHand 3 Ultra, along with the OmniPicker 3 tactile gripper.
The company also highlighted its AGILE motion intelligence framework, which now includes a self-evolving agent capability. It is designed to let robots explore the limits of their physical capabilities and improve through experience in real environments. AGIBOT describes this as a shift from motion control toward spatial mobility intelligence.
GO-3 Preview, WITA-Omni and product launches
AGIBOT debuted GO-3 Preview, a next-generation embodied foundation model built around a framework called WRAM (World Reasoning Action Model). The model is designed to help robots anticipate the consequences of their actions, adapt to changing environments and generalize beyond previously learned task sequences. AGIBOT demonstrated zero-shot and few-shot capabilities and presented its Physical Intelligence Scaling approach, which combines pre-training with post-training and continuous improvement after deployment.
According to evaluation results presented at the event, GO-3 Preview ranked first in both task success rate and process score on RoboDojo and achieved the highest overall score on RoboTwin. The source text is cut off before the remaining benchmark details.
The company also showed its WITA-Omni omni-modal interaction model, with a planned deployment roadmap for more natural, multimodal human-robot interaction, and launched the G2 Max industrial robot and Link-U OS 1.0.
What it means
Worken is a proposal rather than an established standard, and AGIBOT itself says it will be refined with real deployment data. If adopted, it would give buyers and the industry a way to compare deployments by autonomous working time and output, not by unit counts. The company positions the event's announcements as a deployment-first push across hardware, models, operations and ecosystem.
