That's pretty much exactly my train of thought.

I was able to run OLMO 7 billion on my integrated graphics, but I was only getting nine tokens per second output, which is far too low.

I need to actually at some point get my very first dedicated GPU so that I can run an AI model locally.

I'm blind, so I've always just been like, I don't give a shit what GPU is in my computer. I can just use the integrated graphics on the CPU. But that's not the case with AI.

With that said, I will only use fully open source AI models, and as far as I'm aware of, the only fully open source AI model is OLMO, which comes in a 7 billion parameter variant and a 32 billion parameter variant.

I don't think I will need to spend a bunch of Monero on GPUs to run those.

I am actually considering the same thing as my next device as I am very seriously considering using post-market OS as my next operating system.

My RSS list isnt nearly that interesting. My primary use case for RSS is as my YouTube subscription feed since i use NewPipe.

I do have the meshcore blog, the rage, and the postmarket OS blog on it though for news sources.

That's fine, I don't have a brand preference at all. I just need to be able to get it shipped to my house, lol. Of course, I don't buy anything unless it's priced directly in Monero. So there is that.

For other people, yes I can, but I can't buy from myself.