Xiaomi Corp. intensified the competition to develop AI tools for more than text on Monday by releasing an open-source speech model to go along with its home appliance and vehicle technology.
The open-source Qwen2.5-Omni-7B from Alibaba Group Holding Ltd. is integrated with Xiaomi’s fundamental voice model, which has been used in automobiles and smart home devices, to create the new MiDashengLM-7B. In a statement, the Beijing-based phone and car manufacturer described the developments and offered benchmarks.
Beyond its primary smartphone business, Xiaomi has been actively seeking new growth drivers, and electric vehicles are quickly emerging as one of its top business priorities. At the same time, China’s IT industry has made investing in AI development a top priority, and many of the top companies have chosen to publish their work open source in order to attract clients.
In order to better compete with OpenAI’s Sora, major Chinese internet companies, including Alibaba and Tencent Holdings Ltd., have recently released a variety of models that can handle photos, video, and sound. Chinese leader Xi Jinping and US President Donald Trump have both underlined how important it is for their nations to take the lead in the AI race.
