Home / Technology / Xiaomi Unveils Cutting-Edge Open-Source Voice Model for AI Dominance
Xiaomi Unveils Cutting-Edge Open-Source Voice Model for AI Dominance
4 Aug
Summary
- Xiaomi releases new open-source voice model
- Model integrates Alibaba's Qwen2.5-Omni-7B
- Aims to power AI in cars, smart home gadgets

On August 4th, 2025, Xiaomi Corp. has unveiled a groundbreaking open-source voice model, MiDashengLM-7B, to complement its existing automotive and home appliance technologies. The new model is based on Xiaomi's foundational voice system, which has already been deployed in cars and smart home devices.
Notably, the MiDashengLM-7B integrates Alibaba Group Holding Ltd.'s open-source Qwen2.5-Omni-7B, further strengthening Xiaomi's AI capabilities. The Beijing-based company has detailed the advancements and provided benchmarks for the new model in a post on its WeChat account.
This move by Xiaomi signals the company's ambition to push the boundaries of AI beyond just text-based applications. By releasing an open-source voice model, Xiaomi aims to empower developers and manufacturers to incorporate advanced voice recognition and natural language processing into a wide range of products and services.