Calculate precise VRAM requirements for running open-source LLMs (Llama 3, Qwen) with quantization. Compare live cloud GPU rental costs across RunPod, Vast.ai, and Lambda Labs.