GGUF
The standard file format for running AI models on your own computer, packaging the model and its settings into one file.
GGUF is the format you will meet within about five minutes of trying to run a model on your own machine. It bundles everything llama.cpp needs into a single file: the model’s numbers, the vocabulary, and the settings that describe how to use it. Download one file, point the program at it, and you have a working model, no Python environment required.
Most GGUF files you see are quantized, meaning the model’s numbers have been stored with less precision to shrink the file. Names such as Q4_K_M tell you how aggressive that squeeze is. Lower numbers mean a smaller, faster file and slightly worse answers, and Q4 is the usual sweet spot. A 30 billion parameter model that needs 60 GB in full precision often fits under 20 GB as a Q4 GGUF.