Thanks. I’ve been experimenting with local models for over a year now, on and off, so this isn’t just limited to the latest Qwen. Anyway, I have no problem running them, but there’s a huge difference between running something via a chat interface and running it a la Claude Code so that it can interact with the local environment and create/edit files. This is the aspect that’s difficult, in my experience.
It’s all about tooling, if the ai can fetch data it can do something rad with it. Use something like an ai harness to have an mcp server and other tooling to improve the harness and the tools I made this for my own learning: GitHub.com/ralabarge/beigebox
Comments
Thanks. I’ve been experimenting with local models for over a year now, on and off, so this isn’t just limited to the latest Qwen. Anyway, I have no problem running them, but there’s a huge difference between running something via a chat interface and running it a la Claude Code so that it can interact with the local environment and create/edit files. This is the aspect that’s difficult, in my experience.
It’s all about tooling, if the ai can fetch data it can do something rad with it. Use something like an ai harness to have an mcp server and other tooling to improve the harness and the tools I made this for my own learning: GitHub.com/ralabarge/beigebox