I’m seeing a lot of buzz around DeepSeek performance vs cost ratio also seeing claims that it cannot answer questions about tank events and bear cartoons, that’s odd unless specific funding is underreported. I’m not good with language models or geopolitics asking for a friend.