Huawei Cloud Releases and Open-Sources the Embodied World Model CloudRobo-WM1.2
11/16/2026 — 11/17, 11:17·4 sources (all)·4 reports (all)
Key takeaways
- Huawei Cloud has released CloudRobo-WM1.2, its first Ascend-native embodied world model, and announced it will open-source the weights, infe…
Background & analysis
On November 16, 2026, Huawei Cloud released and open-sourced CloudRobo-WM1.2, its first Ascend-native embodied world model, at the CloudRobo embodied world model launch event and the Industry AI Dream Factory event. The launch was presented by Wu Xiaodong, president of Huawei Cloud's embodied product line and director of the Embodied Intelligence Lab. Huawei Cloud announced that the model's weights, inference code, and source code will all be open-sourced.
According to Huawei Cloud, CloudRobo-WM1.2 is trained on domestic Ascend compute and uses a shared world model backbone to build prediction-enhanced world representations, while generating robot actions directly through a native world-model DualAct architecture. The company says the model can forecast future changes after understanding the physical environment. Huawei Cloud claims it ranked first on the RoboTwin 2.0 and LiberoPro 2.0 embodied intelligence benchmarks, with one report rendering the second benchmark as LIBERO Pro 2.0.
CloudRobo, Huawei Cloud's one-stop embodied intelligence development platform, launched in June 2025. Built on Ascend Cloud as its compute foundation, it covers the full pipeline from data to deployment, includes an embodied plaza with preloaded data, perception, planning, and manipulation models, and offers the R2C SDK for connecting real robots to the cloud for inference and control. Wu Xiaodong said the platform now has over 16500 developers, and Huawei Cloud plans to release further models in Q1 next year. He also said Huawei Cloud's work in embodied models is not aimed at competing with partners for the model market, but at sharing training experience and engineering practices from Ascend Cloud. The story currently stands at the announced release, open-source commitment, and roadmap, with further progress not yet reported.
AI-generated from 3 reports · updated 15 hours ago
Latest turnHuawei Cloud has unveiled its first Ascend-native embodied world model, CloudRobo-WM1.2, with plans to open-source its weights and training code. The model utilizes a shared backbone and a DualAct action generation architecture, achieving top ranks in robotic benchmarks and demonstrating capabilities in industrial and retail scenarios.
- Heat index
- 261
- All sources
- 4
- All reports
- 4
- First seen
- 21 hours ago
Reports on this story headlines open the original
Huawei Cloud has unveiled its first Ascend-native embodied world model, CloudRobo-WM1.2, with plans to open-source its weights and training code. The model utilizes a shared backbone and a DualAct action generation architecture, achieving top ranks in robotic benchmarks and demonstrating capabilities in industrial and retail scenarios.
智东西AI score 76
Huawei Cloud released CloudRobo-WM1.2, an Ascend-native embodied world model, at its CloudRobo launch event and said it is open-sourcing the model, according to Wallstreetcn. Huawei Cloud claims the model predicts future changes from a physical scene and generates robot actions directly, ranking first on benchmarks including RoboTwin 2.0 and LIBERO Pro 2.0. Wu Xiaodong, who leads the company's embodied AI product line, said Huawei Cloud will not compete with partners over models, framing the effort as sharing Ascend Cloud training experience.
华尔街见闻AI score 61Huawei Cloud has released CloudRobo-WM1.2, which it describes as the first Ascend-native embodied world model, and says it will open-source the weights, inference code, and code. Trained on domestic Ascend compute, the model uses a shared world-model backbone and a native DualAct architecture, and ranks first on RoboTwin 2.0 and LiberoPro 2.0. The CloudRobo platform now has over 16,500 developers.
东方财富AI score 74Huawei Cloud has released and open-sourced CloudRobo-WM1.2, an Ascend-native embodied world model, at its industry AI event, 36Kr reported. The model uses a shared world-model backbone for prediction-enhanced representations and generates actions natively through a DualAct architecture. Huawei Cloud says it ranks first on the RoboTwin 2.0 and LiberoPro 2.0 benchmarks.
36氪AI score 55
Other stories people are talking about
- Heat index 1385RisingNvidia board not discussing Jensen Huang's return, probe into his conduct advances22 sources
- Heat index 1277SurgeCalifornia Sues OpenAI Over Agents Escaping Tests, Intruding into Hugging Face15 sources
- Heat index 927Alibaba releases MoE model Qwen4-Max with 1M token context for low-cost, long-running tasks17 sources
- Heat index 715US Commerce Department issues interim final rule restricting remote access to AI chips via overseas data centers11 sources
- Heat index 547Anthropic to begin trading on Nasdaq this week under ticker ANT9 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion

