qwen started as the core team of Tongyi dept, which was part of the algorithm & model offerings from Alibaba-Cloud (aliyun)
- qwen series for nlp/LLM
- wan series for cv/video
- fun series for ASR and TTS
- icss for AI customer service
- lingma for coding
- xingchen for metaverse hype
- tingwu for podcast/notebooklm
Now that's the awkward part: qwen was too successful, a team had more influence than Tongyi and even Aliyun, obviously they started to develop multi-modal capabilities overlapping other team's work, even with their own app, very vertical integrated.
But qwen didn't contribute much revenue or DAU/MAU except fame amoung AI communities. Since GPUs were scarce, Alibaba had to balance free open weight models and customer use.
In a better world, alibaba might just as well split qwen into an independent entity and IPO it.