Any option to start running it on local models? Or is that just nowhere near as powerful in terms of output?
For me right now by far not as powerful. But I'm pretty sure that this will change drastically in the next years.