It is amazing how fast it happened. Right now one of my main projects is fully running on DeepSeek flash. My reason was that I was blocked by both of the main US AI labs from working on it because it involves viruses. DeepSeek flash has been killing it since I switched it on, completing the first phase of the project and setting up an iteration in another application space. It isn't the most brilliant model, but it is reliable and I don't have to manage my weekly token allowance. I just spend freely and end up spending only a few dollars a day. Intelligence is going to become a basic commodity. Only special stuff is going to drive us to use special models. And maybe not even that.