llama.cpp ☆

Georgi Gerganov's llama.cpp runs quantized open models on ordinary CPUs, Macs and small GPUs; its GGUF format became the standard for local model distribution. It is the foundation of the local-AI movement and the engine inside Ollama.