The capital required to train Qwen at scale is enormous. The capital required to patch a linux distro is near zero. Any scale model coming out of China should be viewed as advancing a geopolitical agenda. The same skepticism should be applied to any model trained in the states, but that should be viewed through the lens of short-term business profits.

There are earnest nerds everywhere, in every society. No doubt. But "Chinese AI Labs" operate at the whim of the Chinese government, in the same way "American AI Labs" operate at the whim of billionaire investors. Inferring good will from either is naive at best at this scale.