I don’t get the buzz… All that hype and speculation about AI is always very volatile. And almost every new model these days comes with the claim that it outperforms ChatGPT4 albeit being very tiny and super efficient. The publications from Meta and every other western company are worded very similarly… And now someone trained a smaller model on reasoning and published that. But that isn’t a miracle, it could have been done by Mistral or Meta, too… And I’m pretty sure they will. And aren’t AI models getting better and more advanced all the time?