everyone is moving away from expensive subscriptions toward running
small language models locally. it feels much more `
private
` to keep ur data off external servers using ollama run llama3.
it actually works better than i expected if u have enough vram.