AgiBot releases GO-3 Preview, plans open source by end of year
What happened
At the 2026 AgiBot Innovation Day, AgiBot released its next-generation embodied foundation model GO-3 Preview and plans to fully open-source the model and the embodied Agent framework Genie Master by the end of 2026. Co-founder, Co-President, and President of the Embodied Business System Yao Maoqing announced this open-source timeline; whether it can be delivered on schedule will determine whether external developers can actually use it. GO-3 Preview adopts the self-developed WRAM architecture, which unifies task understanding, future state prediction, and action generation into a single model, focusing on full-chain scalable learning across pre-training, post-training, and post-deployment. About 90% of the pre-training data comes from human first-person perspective data. According to AgiBot and multiple reports, the model ranks first on three embodied intelligence benchmarks: RoboDojo, RoboTwin, and RoboColiseum. According to AgiBot, the Genie G2 equipped with Genie Master has been deployed and validated in a real convenience store, continuously improving in tasks such as deli packaging, picking, and restocking without relying on manual demonstrations or debugging. Leikeji reported that its full task process success rate went from 28.6% on the first day to a stable 85.7% after 5 days.
AI 根据报道生成 · 6 days ago更新
Reporting timeline
Follow the reporting to explore different perspectives.
- Robotics & Automation NewsFeaturedAGIBOT releases embodied foundation model GO-3 Preview, tops three embodied AI benchmarks
AGIBOT has released GO-3 Preview, an embodied foundation model built on its self-developed WRAM (World Reasoning Action Model) architecture that unifies task understanding, future state prediction, and action generation. It ranks first on three embodied AI benchmarks: RoboDojo, RoboTwin, and RoboColiseum, and is planned for full open source release by the end of 2026.
- 亿欧网 · 微信公众号Wang Xingxing vs. Zhi Hui Jun: Unitree adopts general models while AgiBot develops in-house models for production lines
The article compares Unitree and AgiBot's divergent approaches to embodied intelligence: Unitree integrates GPT-6 Astra into the G1 as an "external brain," with Wang Xingxing judging that general models will be able to handle planning after another one or two generations; AgiBot only lets Astra serve as System2, while System1 and the 100 to 1000Hz force-control closed loop are developed in-house, and GO-3 Preview tops three leaderboards: RoboDojo, RoboTwin, and RoboCosieum.
- OFweek RoboticsAgiBot Innovation Day releases AHA3, GO-3 and Lingqu OS, proposes Worken work-hour unit
At the 2026 AgiBot Innovation Day on October 29 at the Shanghai World Expo Center, Peng Zhihui released the third-generation hardware architecture AHA3, the foundation model GO-3 Preview, the multimodal interaction model WITA-OMNI and Lingqu OS 1.0, and proposed Worken, a new unit for measuring robot value (number of active robots multiplied by autonomous working hours multiplied by the output efficiency ratio).
- APPSO · 微信公众号Tsinghua-led team lets 96 robots share experience, raising phone packaging success rate to 98.3%; AgiBot releases GO-3 Preview
A Tsinghua-led team enabled 96 robots to share manipulation experience, raising the success rate on a phone packaging task to 98.3%. AgiBot released GO-3 Preview, a unified framework connecting reasoning, prediction, and action.
- 爱范儿Tsinghua-led team lets 96 robots share experience, raising phone-packaging success rate to 98.3%
A Tsinghua-led team let 96 robots share experience, raising the phone-packaging success rate to 98.3%. AgiBot released GO-3 Preview, connecting reasoning, prediction, and action in a unified framework.
- APPSO · 微信公众号Agibot discloses nearly 2 billion yuan in first-half revenue, launches robot "effective intelligent work hours" metric
Agibot disclosed nearly 2 billion yuan in first-half revenue and launched a robot "effective intelligent work hours" metric. The metric is used to measure a robot's actual effective operating time; the original text did not elaborate on the specific statistical scope or verification conditions.
- 爱范儿AgiBot releases GO-3 Preview embodied foundation model, plans to open-source it by year-end
AgiBot released the GO-3 Preview embodied foundation model and plans to open-source it by the end of the year. In the same period, AgiBot disclosed first-half revenue of nearly 2 billion yuan and introduced a "effective intelligent working hours" metric for robots.
- 36Kr Embodied IntelligenceAgibot Releases Embodied Foundation Model GO-3 Preview and Genie Master Plan, Full Open Source by End of 2026
At the 2026 Agibot Innovation Day, Agibot released its next-generation embodied foundation model GO-3 Preview and plans to fully open source the model and the embodied Agent framework Genie Master by the end of 2026. On the same day, Agibot also released Worken, the first "work-hour KPI," introducing KPI assessment standards for robots.
- 36氪 · 微信公众号AgiBot launches "Effective Intelligent Work Hours" Worken, plans to open-source GO-3 Preview and Genie Master by end of 2026
At the 2026 AgiBot Innovation Day, AgiBot launched Worken (Effective Intelligent Work Hours), its first value metric for physical AI, which measures the actual output of autonomous robot work using the number of active robots × autonomous working hours × output efficiency ratio. AgiBot also released its new-generation embodied foundation model GO-3 Preview and plans to fully open-source the model and the embodied Agent framework Genie Master by the end of 2026.
- Robotics & Automation NewsAgibot unveils AHA3 hardware platform, GO-3 Preview model, and G2 Max robot at 2026 Innovation Day
At its October 29, 2026 Innovation Day in Shanghai, Agibot unveiled its third-generation humanoid hardware architecture AHA3, the embodied foundation model GO-3 Preview, the omni-modal interaction model WITA-Omni 1.0, the industrial robot G2 Max, and Link-U OS 1.0, and proposed the Worken framework for measuring robots' actual productive output.
- 36Kr Embodied IntelligenceFeaturedAgiBot Releases GO-3 Preview, WITA-OMNI and AGILE, Connecting the Full Physical AI Chain
At its Innovation Day, AgiBot released the GO-3 Preview embodied foundation model, the WITA-OMNI native multimodal interaction model and the AGILE general motion intelligence algorithm framework, and also launched Lingqu OS 1.0 and the AHA3 third-generation hardware architecture.
- 雷科技AgiBot releases embodied foundation model GO-3 Preview, tops three major leaderboards including RoboDojo
On October 29, AgiBot released its next-generation embodied foundation model GO-3 Preview, which uses the WRAM (World Reasoning Action Model) architecture to connect task reasoning, future prediction, and action generation. It has ranked first on three major evaluation leaderboards: RoboDojo, RoboTwin, and RoboColiseum. AgiBot plans to fully open-source it by the end of 2026.
- 机器之心 · 微信公众号FeaturedAgiBot releases embodied model GO-3 Preview, enabling full-pipeline scaled learning from pretraining to post-deployment
AgiBot has released the embodied model GO-3 Preview, focused on full-pipeline scaled learning across pretraining, post-training, and post-deployment, and it ranked first in three evaluations: RoboDojo, RoboTwin, and RoboColiseum.
- 新智元 · 微信公众号FeaturedAgiBot releases GO-3 Preview embodied model, tops three robotics leaderboards and plans full open source by year-end
AgiBot officially released its GO-3 Preview embodied model on October 29, taking Top 1 on robotics leaderboards including RoboDojo, RoboTwin, and RoboColiseum, and expects to fully open source it by year-end.
- 36Kr Embodied IntelligenceAgiBot releases GO-3 Preview embodied foundation model, plans open source release by year-end and launch of Genie Master framework
At the 2026 AgiBot Innovation Day, Yao Maoqing, AgiBot co-founder, co-president, and president of the embodied business system, officially released the new-generation embodied foundation model GO-3 Preview. AgiBot plans to fully open source GO-3 Preview and the embodied Agent framework Genie Master by the end of 2026.
- 36氪 · 微信公众号AgiBot GO-3 Nano Preview Tops RoboTwin Leaderboard with 85.01% Average Success Rate
AgiBot GO-3 Nano Preview ranks first on the latest RoboTwin leaderboard with an average success rate of 85.01%. The leaderboard evaluates dual-arm collaborative manipulation and environmental generalization in complex scenarios, assessing model performance in target recognition, motion coordination, and adaptation to environmental changes.
Activity for this event
- Current
- 3
- Peak
- 10Nov 5 19:00
- Past 24 hours
- -51%
The trend compares participants observed throughout the same period, so its range may be smaller than the current interest count. Move or tap to inspect each hour; use the arrow keys on a keyboard.