Would it be viable to support GLM 4.6? It's the best open-source coding model we've tried for Shakespeare, but I don't like funneling user data to the Chinese government.