Education & Research Robot Manufacturer Software & Algorithm Provider Human-Machine Interaction
Tencent Open-Sources Three Embodied Foundation Models for Enhanced Robot Performance
Original from PanDaily.com: Tencent Robotics X Open-Sources Three Embodied Foundation Models: Chief Scientist Zhang Zhengyou Explains the 3-Layer Brain Architecture Solving Robot Reaction Speed

Tencent Open-Sources Three Embodied Foundation Models for Enhanced Robot Performance

Tencent has announced the open-source release of three embodied foundation models during the WAIC 2026 event. These models include a Visual Language Model (VLM) designed for scene understanding, the RxBrain cognitive model that facilitates planning with visual states, and the VLA model, which supports continuous action at frequencies between 500 to 1000Hz.

This development is significant as it aims to enhance robot reaction speed and cognitive capabilities, addressing critical challenges in robotic performance. The introduction of these models is expected to advance the field of robotics by providing developers with powerful tools to improve the efficiency and effectiveness of robotic systems.

Looking ahead, the industry will be keen to observe how these open-source models are adopted and integrated into various robotic applications. No further timeline was disclosed at the time of publication.

Editor's Note

The open-sourcing of these foundational models by Tencent reflects a growing trend in the robotics industry towards collaborative development and innovation. By providing access to advanced cognitive and action models, Tencent is positioning itself as a key player in enhancing robotic capabilities, which could influence future technology adoption and competitive dynamics.

RobotToday Initiative

Robotics needs a service framework.

RSF defines a common language for robot service capability, lifecycle operations, certification pathways, and service-provider networks.

Share

Related Articles/News

AgiBot Launches GE-Act 2.0 Native World-Action Model with Enhanced Data Scaling

AgiBot Launches GE-Act 2.0 Native World-Action Model with Enhanced Data Scaling

AgiBot has introduced GE-Act 2.0, a native world-action model that has been pretrained from random initialization using embodied data. This new model has significantly scaled from 300 to 30,000 hours of training, enabling it to perform zero-shot skills, including towel folding, across two different robot embodiments. The release of GE-Act 2.0 is significa...

Household Robots Face Challenges: Market Growth and Open Source Solutions Ahead

Household Robots Face Challenges: Market Growth and Open Source Solutions Ahead

The shipment of humanoid robots is surging, with IDC projecting around 18,000 units globally by 2025, a 508% increase. However, the penetration of these robots into households remains minimal, as most are still used in entertainment, education, and data collection. Ecovacs Group Chairman Qian Dongqi noted that it may take three to five years for humanoid ...

Humanoid Robots Open Source Robotics Household Automation
China’s Dexterous Hand Industry, 2026 Part 2 — The Technology

China’s Dexterous Hand Industry, 2026 Part 2 — The Technology

Linkage, tendon, direct drive or hybrid — and why hardware is converging while the DL1–DL5 manipulation scale decides who wins. Part 2 of 3.

Components and Hardware
Former ByteDance and Tencent AI Researcher Sun Peng Joins Stardust Intelligence for Robot Learning

Former ByteDance and Tencent AI Researcher Sun Peng Joins Stardust Intelligence for Robot Learning

Sun Peng, a former core researcher in AI at ByteDance and Tencent, has joined Stardust Intelligence to improve post-training processes in robot reinforcement learning. This move is expected to enhance the capabilities of robots in learning from their environments more effectively. The significance of this development lies in the growing importance of rein...

Robotics Automation AI
Tencent's Marvis Introduces Custom-Model Feature for Third-Party AI Integration

Tencent's Marvis Introduces Custom-Model Feature for Third-Party AI Integration

On September 1, Tencent launched a new feature for its AI assistant Marvis, allowing users to connect various third-party models such as Kimi and Zhipu GLM. This custom-model capability enables users to run local open-source models and sync them across devices, enhancing the functionality of Marvis. This development is significant as it expands the versat...

ABEJA and Murata Manufacturing Validate Physical AI Technology Using VLA Model for Dual-Arm Robots

ABEJA and Murata Manufacturing Validate Physical AI Technology Using VLA Model for Dual-Arm Robots

On August 31, 2026, ABEJA announced a collaboration with Murata Manufacturing to validate physical AI technology aimed at automating future research and experimental tasks. Utilizing the Vision-Language-Action (VLA) model, they successfully demonstrated a series of coordinated actions, including part transfer, with dual-arm robots in a controlled environm...

Software, Compute and Interaction at WRC 2026

Software, Compute and Interaction at WRC 2026

47 WRC 2026 exhibitors supply robot software, compute or interaction — but only 19 call themselves software companies. Five open artefacts, licence-checked.

Market and Business News
ROBOTTODAY WEEKLY BRIEFING August 24 – 28, 2026

ROBOTTODAY WEEKLY BRIEFING August 24 – 28, 2026

This week in robotics: XPeng's Dogotix raises over $900 million at a $6.3 billion valuation, Unitree's post-IPO slump revives bubble talk, Beijing's World Humanoid Robot Games close with a shift to real-world tasks, and Teradyne sues JAKA at Europe's Unified Patent Court. August 24–28, 2026.

RobotToday Weekly
XPeng Launches Major Upgrade to Second-Gen VLA with Dynamic 4D Spacetime Understanding

XPeng Launches Major Upgrade to Second-Gen VLA with Dynamic 4D Spacetime Understanding

XPeng has introduced a significant upgrade to its second-generation Vehicle Learning Architecture (VLA), enhancing it with dynamic 4D spacetime understanding. This advancement enables production cars to exhibit Robotaxi-grade Level 4 (L4) autonomous behavior, marking a notable step forward in the integration of physical AI. The introduction of 4D spacetim...

Auto
Hugging Face Launches MicroDuck, an Open-Source Robot Amid Nvidia Acquisition Rumors

Hugging Face Launches MicroDuck, an Open-Source Robot Amid Nvidia Acquisition Rumors

Hugging Face has introduced MicroDuck, an innovative open-source robot capable of walking, talking, and roller-skating. Co-founder Thomas Wolf highlighted the increasing demand for robotics and open-source AI, noting that the company is selling approximately one MicroDuck every four seconds. The launch of MicroDuck is significant as it reflects the growin...

Hugging Face Launches $399 Open-Source Microduck Robot Ahead of Christmas

Hugging Face Launches $399 Open-Source Microduck Robot Ahead of Christmas

Hugging Face introduced the Microduck, a $399 open-source duck robot, on Thursday, with shipping expected before Christmas. CEO Clem Delangue described the Microduck as a robot that can be taught new tricks through reinforcement learning, featuring capabilities such as waddling, picking up objects, and roller skating. The Microduck, measuring 25 centimete...

AI Robotics Hugging Face
Comprehensive Survey on Vision-Language-Action Models for Embodied AI

Comprehensive Survey on Vision-Language-Action Models for Embodied AI

A comprehensive survey on vision-language-action models for embodied artificial intelligence has been published in the Journal of Field Robotics. This survey explores the integration of visual perception, language understanding, and action execution in AI systems, highlighting the advancements and challenges in this interdisciplinary field. The significan...

SURVEY ARTICLE

Related Suppliers

Om AI 联汇科技

Om AI (Lianhui Technology) is a Chinese AI firm behind VLX, an edge-deployed multimodal model for perception and decision-making in physical AI hardware.

Roboflow

Iowa-based visual AI platform enabling 1M+ developers to annotate, train, and deploy computer vision models; trusted by half the Fortune 100.

ACCESS

ACCESS Co., Ltd. is a Tokyo-listed (TSE:4813) Japanese software firm known for the NetFront browser, now testing VLA-model remote robot control over IOWN APN.

Humanoid

UK startup building commercially scalable, general-purpose humanoid robots via its KinetIQ VLM/VLA framework, offered as RaaS.

Spirit AI 千寻智能

Beijing embodied-AI startup (founded 2024) building general-purpose humanoid robots and Vision-Language-Action models; deploys Moz1 on CATL lines.

Astribot (Stardust Intelligence) 星尘智能

Shenzhen humanoid robotics firm founded 2022; its cable-driven T1 wheeled humanoid opened orders May 2026 from RMB 89,900, alongside the 10 m/s-arm S1.

KraussMaffei Group GmbH

German machinery group manufacturing injection molding, extrusion, reaction process, and automation systems for plastics and rubber processing since 1838.

Rocket Lab USA 火箭实验室美国

End-to-end space company providing launch services, spacecraft manufacturing, satellite components, and on-orbit management solutions.

Industrial Manufacturing Robot Manufacturer Autonomous Vehicle / UGV

RoboScience 机器科学

Chinese embodied-AI company building cross-embodiment general manipulation foundation models for robots.

GigaAI 极佳科技

Beijing embodied-AI startup building world models (GigaWorld) and a robot brain (GigaBrain) that power humanoid and industrial robots.

PrimeBOT

PrimeBOT, a China-based robotics company, unveiled the Qiyuan T1 in July 2026, calling it the world's first transformable personal humanoid robot.

PNDbotics

Beijing-based full-stack humanoid robot developer, maker of the Adam series (1.67 m, 62 kg, up to 43 DOF) targeting automotive and industrial deployments.

inJoin the RobotToday community on LinkedIn

Daily robotics news, in-depth analysis, conference highlights, and discussions with professionals worldwide.