So there were so many questions coming in, the topic seems to be really hot. Most of the questions were whether the model is good enough to replace the US model. That’s a completely wrong question. I’ll send the benchmarks again: the Deepseek-R1 is the complete model and the Deepseek-R1-32B is what I have on my home computer. So you can see that it is on a par with Open(sic!)AI. The difference, and this is crucial, is that is and under an license. Do what you want with it. Maybe we’ll have it evaluate all the newspaper publications and see which think tank influences which newspaper. Or come up with something of your own.