AGIBOT Proposes Worken Framework to Measure Robot Productivity
At its 2026 Innovation Day, AGIBOT introduced the Worken framework for quantifying robot productivity and previewed its GO-3 AI model, alongside showcasing deployments of its G2 Max industrial robot.
AGIBOT introduced its Worken framework for measuring robot productivity at its Innovation Day 2026 event in Shanghai on October 29. The proposed metric aims to quantify the productive value of physical AI by combining active robot counts, autonomous working hours, and an output-efficiency ratio that compares robot work to a human benchmark. AGIBOT views Worken as an evolving framework that will be refined through real-world deployment data and operational experience. The company's goal is for it to become a broadly applicable reference for the industry.
Worken measures effective intelligent work hours. It counts robots actively performing tasks, the time they spend working autonomously, and their productive output against a task-specific reference benchmark. The framework is still under development. AGIBOT plans to refine it using data gathered from actual deployments.
GO-3 Preview AI Model
AGIBOT also unveiled GO-3 Preview, a next-generation embodied AI foundation model. Presented by Dr. Yao Maoqing, the model is built around WRAM and is designed for reasoning, prediction, and action to help robots adapt beyond pre-learned sequences. It demonstrated zero-shot and few-shot task capabilities. The model employs a Physical Intelligence Scaling approach, which pairs pre-training and post-training with continuous improvement after deployment.
The model achieved top scores on several industry benchmarks. Its performance is detailed in the table below.
G2 Max Robot Deployment
The G2 Max, a heavy-duty industrial robot, is now deployed across multiple manufacturing and logistics sectors. The robot features 21 degrees of freedom and a dual-arm handling capacity of up to 40 kilograms. Its working height is adjustable from 0.1 to 2.5 meters.
Deployments are underway in varied applications. These include tablet production at Longcheer, semiconductor packaging and testing, CNC machine tending, and logistics operations. The robot is also used at the Guangzhou Metro and the Chimelong tourism complex. Its applications extend to industrial material transport, warehouse handling, and pallet-related work.
Industry Innovation Context
The 2026 Embodied Intelligence Industry Innovation Report highlighted Worken and AGIBOT's deployment examples as key industry benchmarks. The report was jointly compiled by the China Economic Information Service and the Shanghai Industrial Innovation Center of CAICT. AGIBOT's broader robotics ecosystem includes 255 universities, 511 companies, and more than 80,000 developers.
Alongside the major announcements, AGIBOT launched several other products. The company unveiled the AHA3 integrated hardware platform, which unifies locomotion, interaction, and manipulation intelligence. Its Link-U OS 1.0, claimed to be the industry's first operating system purpose-built for embodied intelligence, connects robot hardware with upper-layer applications and is being trialled by selected partners. The system will become fully open-source by the end of 2026.
The WITA-Omni 1.0 supports full-duplex conversation with coordinated speech and movement for on-device deployment. For manipulation, AGIBOT expanded its tactile offerings with the OmniHand 3 Lite, OmniHand 3, OmniHand 3 Ultra, and the OmniPicker 3 tactile gripper. The company also launched the MEgo Glove, a wearable tactile data-collection device offered in Air and Ultra versions, with the Ultra model adding 3D fingertip and palm force data. AGIBOT's portfolio spans humanoid robots, quadrupeds, dexterous systems, and commercial cleaning systems. The company plans to refine the Worken framework using real-world deployment data and make its Link-U OS fully open source by the end of 2026.
