llama.cpp

license

MIT(OSI)

source

public

core features withheld

no

Fully MIT-licensed C/C++ inference engine: the entire source is public and there is no proprietary tier. Neither the repository nor the ggml-org GitHub org profile offers a hosted or paid SKU, so the only distribution is the one MIT-licensed codebase.

• https://github.com/ggml-org/llama.cpp recorded 2026-09-22

  Public repository, single MIT license badge, README states "Plain C/C++ implementation without any dependencies"; no pricing page, paid tier, or feature-gated build found in the repo

• https://github.com/ggml-org/llama.cpp/blob/master/LICENSE recorded 2026-09-22

  MIT License text, "Copyright (c) 2023-2026 The ggml authors"

No download or user count is published for llama.cpp, so its adoption is read from its standing in local inference. Hugging Face, which ggml.ai joined in February 2026, calls it the fundamental building block for local inference, and the Hub documents GGUF usage with llama.cpp beside Ollama, LM Studio and GPT4All. Neither source says how many people use it.

• https://huggingface.co/blog/ggml-joins-hf recorded 2026-10-09

  "GGML and llama.cpp join HF to ensure the long-term progress of Local AI" (20 February 2026), stating "llama.cpp is the fundamental building block for local inference"; no usage, download or user figure

• https://huggingface.co/docs/hub/gguf-llamacpp recorded 2026-10-09

  Separate Hub doc pages titled "GGUF usage with llama.cpp", "GGUF usage with LM Studio", "GGUF usage with GPT4All", and "GGUF usage with Ollama"

llama.cpp is the frontier for portable, CPU and consumer-hardware inference and for quantization breadth, rather than the datacenter-GPU throughput frontier occupied by vLLM, SGLang and TensorRT-LLM, which is what keeps it a step below them.

• https://github.com/ggml-org/llama.cpp recorded 2026-08-09

  Backend list (Metal/CUDA/HIP/MUSA/Vulkan/SYCL/WebGPU/OpenVINO), 1.5-8bit quantization list, multimodal subsystem reference, "llama-server REST API", OpenAI-compatible server example

• https://raw.githubusercontent.com/ggml-org/llama.cpp/master/docs/models.md recorded 2026-09-29

  The models doc says "The [Hugging Face](https://huggingface.co) platform hosts [thousands of models]" compatible with llama.cpp, and requires "the model to be stored in the [GGUF](https://github.com/ggml-org/ggml/blob/master/docs/gguf.md) file format".

• https://raw.githubusercontent.com/ggml-org/llama.cpp/master/README.md recorded 2026-09-29

  README backend table lists "Metal", "CUDA", "HIP", "Vulkan", "SYCL", "WebGPU" and "OpenVINO [In Progress]" backends; the description lists "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"; it links the "llama-server REST API", offers "Launch OpenAI-compatible API server" and credits libraries "used by multimodal subsystem".