All anyone ever talks in this sub is about x new model having x benchmark scores and how well it can code. Or some new tech to increase t/s for qwen. So I wanted to go back to the roots of this sub and talk about how well local models can write. I am not going to mention closed source models, this d
