DeepSeek R1 just dropped – but don't ask it about Tiananmen Square

If you download the model from Huggingface with a tool like Ollama, deepseek-r1 will answer these types of questions

Not sure if it is the official web frontend doing this or if they are running different models than were uploaded to HF

I vaguely remember something similar when querying google's gemini models about yt-dl. We need a diverse ecosystem quickly to combat this.