# ggml-org/llama.cpp

LLM inference in C/C++

Repository: https://github.com/ggml-org/llama.cpp
Canonical: https://ross.abutalabs.com/products/llamacpp
Homepage: https://llama.app
Language: C++
License: MIT
License Family: permissive
Topics: ggml
Last push: 2026-08-26T22:00:32+00:00

## Health v2 (maintenance only)
Score: 93/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 87, longevity 90
- inputs: {"age_days": 1272, "days_push": 7, "days_rel": 8, "gap_med": 0.0, "n_releases_24m": 4641}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 125803, forks 22236 (observed 2026-08-28T04:12:24.174895+00:00)

## What it is
llama.cpp is a C/C++ library and toolset for running large language model inference locally on CPUs and GPUs, built on the ggml tensor library. It ships CLI tools, an OpenAI-compatible REST server (llama-server), and supports quantized GGUF models across a wide range of hardware.

## Use cases
- run an LLM locally on my laptop without an API key
- serve a local model with an OpenAI-compatible API
- run quantized GGUF models on CPU
- run LLM inference on Apple Silicon
- host a private chatbot on my own machine
- run a local coding agent with a local model
- run LLM inference on a Jetson or embedded GPU

## When to choose
- you want private, fully local LLM inference with no telemetry or API keys
- you need to run models on heterogeneous hardware from laptops to clusters
- you want a lightweight C/C++ inference stack with quantization support

## When to avoid
- you need large-scale multi-node training or fine-tuning infrastructure
- you prefer managed cloud inference APIs
- you need a high-level Python training framework rather than an inference runtime

## Facets
- artifact type: library
- maturity: stable
- function: llm-inference, machine-learning, cli, http-server, gpu-computing
- domain: large-language-models, artificial-intelligence, machine-learning, developer-tools, self-hosted
- platform: cross-platform, windows, cli, wasm, python, cpp, c
- tags: ggml, local-llm, quantization, cpu-inference, apple-silicon, openai-compatible-api, llama-server, linux, macos, docker, gpu

## Member repositories
- ggml-org/llama.cpp (main) score 93

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:12:24.174895+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:11:31.312590+00:00, confidence not recorded.
  - readme: https://github.com/ggml-org/llama.cpp (fetched 2026-08-28T04:12:24.174895+00:00, sha 2c343a497046)
  - homepage: https://llama.app (fetched 2026-08-28T17:27:56.655614+00:00, sha 003dcf48936b)
  - site_page: https://llama.app/docs (fetched 2026-08-28T17:27:56.664125+00:00, sha 44136fa355b3)
- Data as of 2026-08-30T08:39:29.467469+00:00.