Interesting.

Chatting GLM-5(reasoning)(preview) answers just fine. Even after restricting web search and giving answer based on its own knowledge. Probably the results were different in China?

(GLM-4.7 failed to know anything without web search)