Chinese AI Companies Shift to Proprietary Models for Commercial Benefits
Major Chinese artificial intelligence firms are increasingly developing proprietary AI models rather than releasing them as open-source software. This week, Alibaba introduced three new proprietary models that are exclusively available through its official cloud platform and chatbot website. The models include Qwen3.6-Plus, which features improved coding capabilities, and Qwen3.5-Omni, a multimodal system capable of processing text, audio, images, and video. Notably, the previous version of the Omni model, Qwen3-Omni released in September, was made open-source, but the latest iteration will remain proprietary. According to an Alibaba Cloud representative, the decision not to open-source the new Omni model is based on lower developer interest, as indicated by download statistics on the Hugging Face platform. While companies like Alibaba Cloud and Zhipu AI have not abandoned open-source strategies entirely, this development reflects a broader industry trend where advanced models are becoming larger and more complex, making them difficult to run on local hardware. This shift allows companies to capture full commercial value through revenue-generating channels. Alibaba Cloud is the AI and cloud computing division of Alibaba Group Holding.
Key facts
- Alibaba released three proprietary AI models this week
- Models include Qwen3.6-Plus with enhanced coding abilities
- Qwen3.5-Omni is a multimodal model processing text, audio, images, video
- Previous Omni model (Qwen3-Omni) was open-sourced in September
- Latest Omni model will not be open-sourced due to lower developer interest
- Models are accessible only via Alibaba's official cloud platform or chatbot website
- Decision reflects industry trend toward proprietary models for revenue generation
- Advanced models are growing in size, making local hosting difficult
Entities
Institutions
- Alibaba Cloud
- Alibaba Group Holding
- Zhipu AI
- Hugging Face
Locations
- China