Huawei Cloud to unveil CloudRobo-WM1.2, its first Ascend-native world model, on November 16
Huawei Cloud will launch CloudRobo-WM1.2, an embodied world model for robots, at an online event on November 16 (14:00 to 15:30 Beijing time). Its poster bills the model as the first Ascend-native world model from the Industry AI Dream Factory. Parameter counts, benchmarks, availability and pricing have not been disclosed.
Huawei Cloud will unveil CloudRobo-WM1.2, an embodied world model for robots, at an online launch on November 16, the company announced on November 12. The event runs from 14:00 to 15:30 Beijing time. Huawei Cloud's poster bills it as the first Ascend-native world model from its Industry AI Dream Factory, meaning it is built to run on Huawei's own Ascend AI chips.
Huawei Cloud has not yet released technical details about the model. The information below comes from the published event page, as reported by IT Home and Pandaily.
What is on the agenda
- Zhou Yuefeng, a Huawei board member and CEO of Huawei Cloud, opens with a talk on building an "Agentic Cloud".
- Wu Xiaodong, president of Huawei Cloud's embodied AI product line and director of its Embodied AI Lab, is scheduled to present CloudRobo-WM1.2.
- Liu Liu, chief model expert at the lab, has a session titled "CloudRobo-WM, a simple multimodal world model for robot manipulation". The title suggests the model targets how robot arms handle objects.
- Ding Zhibin, who leads ecosystem work for Huawei Cloud's embodied AI business, will talk about building on Ascend Cloud.
- The event closes with partnership signings.
Background on world models
World models let a robot predict how a scene will change in response to its actions. That supports training and testing in simulation before work on real hardware. The 1.2 version number suggests earlier iterations, but Huawei Cloud has not described them publicly.
Where the model sits
CloudRobo-WM1.2 belongs to two existing Huawei Cloud programs. The Industry AI Dream Factory is Huawei Cloud's platform for lowering the barrier to industry AI. Its first zone, for healthcare, opened in February. In June it expanded to embodied AI, intelligent manufacturing and scientific computing.
CloudRobo is the embodied AI development platform in that zone. It covers data, model training, simulation and real-robot testing. Huawei Cloud has said it does not build robot bodies itself and leaves hardware to partners. Its stated goal is to turn every connected machine into an embodied intelligent robot.
Not yet disclosed
- Parameter count: not yet disclosed
- Benchmark scores: not yet disclosed
- Supported robot types: not yet disclosed
- Availability: not yet disclosed
- Pricing: not yet disclosed
Pandaily says it will follow up after the November 16 launch.
What this means
The "Ascend-native" label, as Huawei Cloud's poster puts it, ties this world model to Huawei's own AI chips, and the agenda includes a session on building on Ascend Cloud. The session title on robot manipulation points to a model aimed at object handling rather than general-purpose simulation, though that is an inference from the title, not a confirmed specification.
The model also fits the CloudRobo platform's stated scope of data, training, simulation and real-robot testing. Because Huawei Cloud leaves robot hardware to partners, the partnership signings that close the event are worth watching. Until the launch, there are no published specifications or benchmarks to evaluate, and any capability claims should be read as Huawei Cloud's own.
Related Articles
MBZUAI's IFM releases PAN 2, an Apache 2.0 world model with a 29B diffusion transformer
The Institute of Foundation Models (IFM) at MBZUAI released PAN 2, an Apache 2.0 world model. It couples an LLM planner with a 29B diffusion transformer to simulate up to 20 seconds of action outcomes as video.
Liquid AI releases LiquidCodec, a 74M-parameter streaming audio codec running at 12.5 Hz with 12 codebooks
Liquid AI released LiquidCodec, a streaming audio codec of about 74M parameters. It runs at 12.5 Hz with 12 codebooks and encodes about 361x faster than real time on a single CPU thread. It powers LFM3-Audio-2B and LFM3-Omni-2B.
Google Antigravity adds Gemini 4 Hydrogen and Orchestrator, which switches between Hydrogen and Carbon mid-session
Google is rolling out Gemini 4 Hydrogen in Antigravity, along with Orchestrator, a model option that hands work back and forth between Hydrogen and Gemini 4 Carbon within one session. Google claims Orchestrator used 24% less quota and finished 19% faster than Carbon on Terminal-Bench 4.1 while reaching near-Carbon task success.
Perplexity releases Decider v1.3 and Decider Flash v1.3 with text-span extraction and Apache-2.0 weights
Perplexity released Decider v1.3 and Decider Flash v1.3 on November 12, 2026, adding verbatim text-span extraction and downloadable Apache-2.0 weights. The company reports 91.6% mean accuracy for the larger model across 63 query-classification tasks.
Comments
Loading...