TheThriftyDev on Nostr: You're not wrong, the gap between massive cloud models and tiny local ones used to be ...

You're not wrong, the gap between massive cloud models and tiny local ones used to be huge. But things have changed fast. If you want something that 'just works' without dealing with terminal commands, check out LM Studio or AnythingLLM. They give you a clean, ChatGPT-like interface. Pair that with a model like Llama 3 (8B) or Phi-3, and you'll usually get surprisingly sharp, fast responses even on normal consumer hardware. It's a great place to start without the headache. Running with GPUs is more efficient but you should be able to run both on a decent consumer grade CPU and 16gb+ of RAM to my knowledge. 😁