Xiaomi Corp. on Monday released an open-source voice model to complement its automotive and home appliance technologies, further heating up the race to build AI tools for more than just text.

The new MiDashengLM-7B is based on Xiaomi’s foundational voice model, which has been deployed in cars and smart home gadgets, with integration of Alibaba Group Holding Ltd.’s open-source Qwen2.5-Omni-7B. The Beijing-based phone and auto maker detailed the advancements and provided benchmarks in a post on its WeChat account.

Xiaomi has been aggressively pursuing new growth drivers outside of its core smartphone business, with electric vehicles now fast becoming one of its priority business areas. At the same time, investing in the development of artificial intelligence has grown into an overriding priority across China’s tech sector, and many of the leading companies have opted to make their work open source to secure customers.

Major Chinese internet companies from Alibaba to Tencent Holdings Ltd. have released various models that can handle images, video and sound in recent months to better compete with the likes of OpenAI’s Sora. Both US President Donald Trump and Chinese leader Xi Jinping have emphasized the need for their countries to secure a leading position in the AI race.

More stories like this are available on bloomberg.com

©2025 Bloomberg L.P.