We do, but the vast majority of users interact with centralised models from Open AI, Google Gemini, Grok...

I use gpt4all and have downloaded some models. It's not that hard, and doesn't take a huge system. It's pretty easy if you use models on their list, it's a bit more work to find the right chat interactive script (forgot what they are called) for your class of model if you downloaded one that's not on the list, but not that hard. I have 8GB VRAM, and it's slow but they work. I like that it's local and private, but then I don't use them for much other than as an oddity that is fun to interact with.