2025 CES文章顶部

Huawei Launches Large Model-Powered HarmonyOS 4, Escalating On-Device AI Race

Underpinned by Huawei's pre-trained deep learning AI model Pangu Models 3.0, the voice assistant Xiaoyi can offering more intelligent experience to HarmonyOS 4.

BEIJING, August 7 (TMTPost)—Huawei launched its latest version of its operating system HarmonyOS 4.0 in Huawei Developer Conference (HDC) 2023, escalating tech sector’s race in on-device artificial intelligence (AI).

Credit:Huawei

Credit:Huawei

Compared with the previous version, HarmonyOS 4.0 gets 20% more smooth operation and increases battery life by 30 minutes. It has a key feature with the updated Xiaoyi, Huawei’s voice assistant also known as Celia. Underpinned by Pangu Models 3.0, the latest version of its Pangu pre-trained deep learning AI model, Xiaoyi can access to large AI model, offering more intelligent experience to the system, announced Yu Chengdong, or Richard Yu, CEO of Huawei Consumer Business Group as well as CEO of Huawei Intelligent Automotive Solution Business Group.

The development of large AI model technology will bring about the smart experience of the next-generation intelligent terminal operating system, Yu said at the annual HDC event. With underlying capabilities of Pangu Models 3.0, Huawei will bring users a brand-new AI experience revolution in smart terminal interaction, high-level productivity efficiency, and personalized services. The telecom giant also vows to create the more responsible AI and make identifying all the AI-generated content as it advocates the security-and-privacy-first approach in AI experience innovation and scenario design.

Like other applications back by famous rival large models, HarmonyOS 4.0 also can recognize the content and text in the picture, read the text and access more services through Xiaoyi. Xiaoyi can automatically complete specific tasks if users talk to it like daily conversation with other people, just as iPhone’s old feature with help of Siri.

However, as the first intelligent assistant backed by large AI model, Xiaoyi has a deeper understanding of natural semantics. It can understand less obvious voice commands, so smartphones equipped with HarmonyOS 4 can complete tasks more accurately and quickly than iPhone. In addition, Xiaoyi can access services faster since it has AI generalization capabilities and mobile smart assistants in connection with a variety of services and special scenarios. For instance, if a mobile user gives instructions regarding to an event poster sent to Xiaoyi, it can automatically recognize the address on the picture, and give navigation buttons or save the phone number of the poster as the contact information.

Huawei has started rollout of HarmonyOS 4.0’s first public beta for 34 devices such as Mate 50 series, P60 series, Mat X3, MatePad Pro last Friday. The company will updated the system for more devices such as smart watches, smart displays and routers in the second half of the year. Yu said there are 700 million units of smart devices that have been equipped with HarmonyOS, ranging from smartphones, tablets, smart displays and car cabins, and HarmonyOS developers have exceeded 2.2 million.

As the center of HarmonyOS ecosystem, Huawei’s smartphone business is set to get more traction following wide appplication of HarmonyOS 4.0 backed by Pangu Models.

Unlike the mainstream large models such as ChatGPT, Pangu Models 3.0 doesn’t know how to compose poems, but it can do concrete works with three purposes for innovation—reshaping industries, honing technologies and sharing success, to continue to build core competitiveness and serve customers, partners in the industry and developers with better services, said Zhang Pingan, executive director of Huawei and CEO of Huawei Cloud, when rolling out the model.

As an industry-oriented large model, Pangu Models 3.0 is made up of three layers with so-called “5+ N+ X” architecture. The foundation layer, L0, represents the “5”. It provides various skills to meet the needs of different industry scenarios with its five foundational models in natural language processing, computer vision, multimodal learning, prediction and scientific computing. Pangu Models 3.0 is available in serialized foundational models with four different parameters numbers including 10 billion, 38 billion, 71 billion and 100 billion, to satisfy customers’ diversified industry demands in terms of scenario, latency, and response speed.

转载请注明出处、作者和本文链接
声明:文章内容仅供参考、交流、学习、不构成投资建议。
想和千万钛媒体用户分享你的新奇观点和发现,点击这里投稿 。创业或融资寻求报道,点击这里

敬原创,有钛度,得赞赏

赞赏支持
发表评论
0 / 300

根据《网络安全法》实名制要求,请绑定手机号后发表评论

登录后输入评论内容

扫描下载App