Sign inCreate account
Sign inCreate account

llama.cpp

From public sources

llama.cpp is an MIT-licensed C/C++ LLM inference engine that runs large language and vision models locally or in the cloud with minimal setup and no dependencies.

AI toolCategory: AI ToolsOpen sourcedata checked Oct 15, 2026
Add to compareAdd to collection

Overview

Running large language models usually means depending on heavy frameworks, cloud APIs or specialized hardware. llama.cpp is an open-source LLM and VLM inference engine written in plain C and C++ with no external dependencies, designed to run models with minimal setup and strong performance across a wide range of hardware. It supports quantization from 1.5-bit to 8-bit integers to reduce memory use, and provides optimized backends for Apple silicon via Metal, x86 AVX and AMX, NVIDIA and AMD GPUs through CUDA and HIP, plus Vulkan, SYCL and other targets. It can split models across CPU and GPU when they exceed available VRAM. Models can be downloaded directly from Hugging Face and run through the llama CLI or the llama-server OpenAI-compatible API. It suits developers, researchers and hobbyists who want to run models locally, self-host an API, or deploy inference on their own servers. Installation is available via install scripts, Docker, pre-built binaries or building from source. The project is released under the MIT license.
  • ggml
  • llm
  • local-llm

Installation

Commands are shown, never run by the marketplace. Check them against the official documentation.
  • Official websiteFrom public sourcesfrom catalog metadata
    llama.app/ ↗
  • Language package registrypythonFrom public sourcesfrom the package manifest — publication on the registry is not verified
    pipx install llama-cpp-scripts

    Not verified: check that this package in the registry belongs to this project before installing.

    pypi.org/project/llama-cpp-scripts/ ↗
  • Build from sourceFrom public sourcesfrom catalog metadata

    Build from source: no validated build recipe — follow the repository instructions.

    github.com/ggml-org/llama.cpp ↗

Compatibility

Unknown means no evidence was found — not that the product is incompatible.
PlatformStatusArchitectureEvidenceConfidence
WindowsUnknown———
macOSUnknown———
LinuxUnknown———
Web browserUnknown———
Server / self-hostedUnknown———
Container (OCI)Unknown———
PythonCompatible—package manifest: pyproject.toml name0.60

Releases

checked Oct 29, 2026
v0.8.0LatestOct 28, 2026Official source ↗

Files in 0.8.0

The marketplace does not host these files: downloads go to the official source. Files have not been scanned by us.
FilePlatformSizeSHA-256Download from the official source
nightly-tag.txtother7 B724adc02ee64…Official source ↗

Release history

Quality profile

algorithm quality-v1.0 · assessed Oct 15, 2026
Quality Score91
Risk Score0
Confidence 1.00
Quality Score and Risk Score are different measures. A high score does not guarantee safety.

Score components

Maintenance · 2020
last activity 0 days ago
Documentation · 1515
README 7470 chars, description present
Ready to use · 156
install instructions: no; releases: yes
Metadata · 1010
4/4 of description, homepage, license, topics
Releases · 1010
10 release(s), latest 0 days ago
Category confidence · 1010
classifier confidence 1
Adoption · 1010
131301 stars (log-dampened, 10k = max)
License clarity · 55
SPDX MIT
Community · 55
issues enabled, 24441 forks

License

MIT
Known licensefrom the repository host

This is the catalog policy status, not legal advice.

Sources

github.com/ggml-org/llama.cppPrimary★ 131,301 starslast activity Oct 15, 2026Primary repository · Confidence 1.00

Reviews

Sign in to write a review

No reviews yet.

Alternatives

ollamaSimilar
Ollama is an open-source AI tool for running open models such as Gemma, DeepSeek, Qwen and gpt-oss locally or in the cloud, with a CLI, REST API and Python and JavaScript libraries.MIT · AI Tools
SoupSimilar
Fine-tune LLMs from one YAML. Layer streaming trains an 8B model on a 4 GB laptop GPU.Apache-2.0 · AI Tools
haxSimilar
A minimalist, terminal-native coding agent written in C.MIT · AI Tools
selfhost-aiSimilar
selfhost-ai is a Docker Compose installer that deploys 30+ open-source AI and automation tools, such as n8n, Ollama and Open WebUI, on your own server with one command.Apache-2.0 · AI Tools
janSimilar
Jan is an open-source AI tool and ChatGPT alternative that runs large language models locally on your computer, fully offline, for private chat and assistants.AI Tools
code2promptSimilar
A CLI tool to convert your codebase into a single LLM prompt with source tree, prompt templating, and token counting.MIT · AI Tools

README

From ggml-org/llama.cpp · license MIT · fetched Oct 29, 2026

llama.cpp

[llama]

LLM inference in C/C++

[License: MIT] [Release] [Nightly] [Server] [Docker] [Winget]

ggml / ops / maintainer PRs / dev stats / lib llama API / llama-server REST API

Quick start

A few options to get llama.cpp installed on your machine:

# curl
curl -LsSf https://llama.app/install.sh | sh

# powershell
irm https://llama.app/install.ps1 | iex

Once installed:

# Download and run a model directly from Hugging Face
llama cli -hf ggml-org/Qwen3.5-0.8B-GGUF

# Launch OpenAI-compatible API server
llama serve -hf ggml-org/Qwen3.5-0.8B-GGUF

Description

The main goal of llama.cpp is to enable LLM (and VLM) inference with minimal setup and state-of-the-art performance on a wide range of hardware - locally and in the cloud.

  • Plain C/C++ implementation without any dependencies
  • Apple silicon is a first-class citizen - optimized via ARM NEON, Accelerate and Metal frameworks
  • AVX, AVX2, AVX512 and AMX support for x86 architectures
  • RVV, ZVFH, ZFH, ZICBOP and ZIHINTPAUSE support for RISC-V architectures
  • 1.5-bit, 2-bit, 3-bit, 4-bit, 5-bit, 6-bit, and 8-bit integer quantization for faster inference and reduced memory use
  • Custom CUDA kernels for running LLMs on NVIDIA GPUs (support for AMD GPUs via HIP and Moore Threads GPUs via MUSA)
  • Vulkan and SYCL backend support
  • CPU+GPU hybrid inference to partially accelerate models larger than the total VRAM capacity

The llama.cpp project is build on top of the ggml library.

Supported backends

BackendTarget devices
BLASAll
BLISAll
CANNAscend NPU
CUDANvidia GPU
HIPAMD GPU
HexagonSnapdragon
IBM zDNNIBM Z & LinuxONE
MUSAMoore Threads GPU
MetalApple Silicon
OpenCLAdreno GPU
OpenVINO [In Progress]Intel CPUs, GPUs, and NPUs
RPCAll
SYCLIntel GPU
VirtGPUVirtGPU APIR
VulkanGPU
WebGPUAll
ZenDNNAMD CPU

Documentation

Tools
Development

Contributing

  • Contributors can open PRs
  • Collaborators will be invited based on contributions
  • Maintainers can push to branches in the llama.cpp repo and merge PRs into the master branch
  • Any help with managing issues, PRs and projects is very appreciated!
  • Read the CONTRIBUTING.md for more information

Acknowledgements

  • yhirose/cpp-httplib - Single-header HTTP server, used by llama-server - MIT license
  • nothings/stb - Single-header image format decoder, used by multimodal subsystem - Public domain
  • nlohmann/json - Single-header JSON library, used by various tools/examples - MIT License
  • mackron/miniaudio - Single-header audio format decoder, used by multimodal subsystem - Public domain
  • sheredom/subprocess.h - Single-header process launching solution for C and C++ - Public domain
Open the repository ↗