llama.cpp — Ultimate Local LLM Inference Engine
75K+ stars — run any AI model offline on any computer, the backbone of local AI
The most popular open-source project for running AI language models locally on any hardware. Pure C/C++ implementation with no dependencies. Runs Llama, Mistral, Phi, Gemma, and hundreds of other AI models on CPU, GPU, or Apple Silicon. 75,000+ GitHub stars. Supports GGUF format for efficient offline inference. When cloud AI dies, llama.cpp keeps your local intelligence alive. The backbone of offline AI worldwide.
Open llama.cpp — Ultimate Local LLM Inference Engine →
Tags
local LLM · offline AI · C++ · inference engine · GGUF · CPU inference · open source · Llama · Mistral · no internet AI · self-hosted · apocalypse ready