Blaster M

Most of them are, and on github

jan, ollama, kobold, there’s a few out there for jibber jabber. There’s also comfyui, invoke, sdnext, for images and video, and I’m still looking into the audio ones

Local models. Can’t be surveiled if your ai isn’t on the internet.

I remember Mythbusters trying this, and while other factors probably influenced more, they seem to have responded to death metal the best.

Just a bit stiff