Hello, and welcome to impressive local LLM's

I have been working with a few things lately and one of them has been local LLM models for coding. I have a guide here to help set up your own solution if you are interested. But the reason I am posting today is just to discuss the giant leap that is Qwen 3.8 27B. I saw some posts about that and decided to update my local model to it. Now I have heard its not as good at general chat and knowledge, I haven't really tested that. I have tested it with Hermes and VScode, and let me tell you what...It's pretty dang good.

It plugs away at tasks, creates todos, runs tests, uses tools well. I have watched it lean into an ask for well over an hour, working and compressing context, and able to be steered, and working some more. I used to to make imporvements to the fellowship steps tool, and add a kanban board tool, and a few others. All over the course of a couple of days.

I am genuinly impressed at the ability to create something (even if small and there are hundreds or thousands of others out there) for myself that works and is custom and realitivly simple. While I don't think there is any real human intelegence in these tools, I do think we have moved beyond the "this is wordpress or autocorrect on steroids". Depending on your goals and tasks you are trying to accomplish these tools are getting genuinely useful.