Meta's open-weight model family — run locally, fine-tune freely, and deploy without API dependency.
Five passes over the same idea, each from a different angle. Do them in order, or jump to whichever you need.
Llama (Meta) is the most popular open-weight model family. Llama 3 and beyond offer competitive quality with full control: self-host on your infrastructure, fine-tune on your data, and avoid vendor lock-in. Key considerations include quantization (GGUF, GPTQ), inference servers (vLLM, llama.cpp, TGI), and hardware requirements.
