Collection · A local stack, top to bottom

Nothing leaves the laptop

7 picks · 2 of the atlas's 14 topics · 693,386 combined stars · snapshot 2026-10-22

Sometimes the requirement is not cost or latency but that the code, the documents and the prompts stay on hardware you own. That rules out most of the atlas. What is left is a genuine stack: an engine, a runner in front of it, a chat surface, a serving layer for when one machine is not enough, and retrieval over your own files. Every project below names a local runtime as a target, and the build checks it.

Open all 7 in the atlas →

From there: save them to your own projects, or export the set as Markdown, a standalone HTML page, or a PDF.

Runs on: Windows WSL2 macOS Linux Docker

  1. 1 The runner

    Ollama

    ollama/ollama

    Pull a model, run it, done. It is the layer that made local models an afternoon rather than a weekend, and it stays out of the way afterwards.

    Ollama is a tool for running large language models locally, offering easy setup for macOS, Windows, Linux, and Docker, along with a library of models and quickstart guides for customization and integration github | github profile

    irm https://ollama.com/install.ps1 | iex

    184,208 stars · 4 of 39 lists · Go · MIT · pushed 2026-10-22

    Windows WSL2 macOS Linux Docker

    Harnesses & Runtime Infra Local / Ollama

  2. 2 The engine

    llama.cpp

    ggml-org/llama.cpp

    CPU, GPU and Apple Silicon inference in C++, and the foundation most of the rest of this page is standing on. Quantisation happens here, which is where the speed comes from.

    C/C++ inference. CPU, GPU, Apple Silicon. Foundation of local AI

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

    131,944 stars · 3 of 39 lists · C++ · MIT · pushed 2026-10-21

    Windows WSL2 macOS Linux Docker

    Harnesses & Runtime Infra Local / Ollama

  3. 3 The zero-setup option

    Llamafile

    mozilla-ai/llamafile

    A model and its runtime as one executable file. Nothing to install and nothing to uninstall, which makes it the right thing to hand to somebody else.

    Mozilla project that packages LLMs into single executable files that run anywhere

    git clone https://github.com/Mozilla-Ocho/llamafile.git

    26,452 stars · 2 of 39 lists · C++ · no licence stated · pushed 2026-10-12

    Windows WSL2 macOS Linux Docker

    Harnesses & Runtime Infra Local / Ollama

  4. 4 The chat surface

    Open WebUI

    open-webui/open-webui

    A self-hosted interface with access control and extensions, so the local stack has a front door other people can be given a key to.

    User-friendly AI Interface (Supports Ollama, OpenAI API, ...)

    pip install open-webui

    156,376 stars · 2 of 39 lists · Python · no licence stated · pushed 2026-10-21

    Windows WSL2 macOS Linux Docker

    Harnesses & Runtime Infra Codex / OpenAILocal / Ollama

  5. 5 The consumer-hardware path

    GPT4All

    nomic-ai/gpt4all

    Built for the machine you already have rather than the one you would need. Worth trying first if the honest answer about your GPU is that there isn't one.

    OSS local chat. Consumer hardware

    pip install gpt4all

    77,351 stars · 1 list · C++ · MIT · pushed 2025-05-27

    Windows WSL2 macOS Linux Docker

    Harnesses & Runtime Infra Local / Ollama

  6. 6 Your own documents

    Local GPT

    PromtEngineer/localGPT

    Retrieval over local files with a local model doing the reading, which is the whole point: the documents never become somebody else's training data. Two source lists carry it.

    LocalGPT is an open-source project for secure, private interactions with documents locally, featuring comprehensive model support, embeddings, API for RAG applications, and GUI options, with a focus on privacy and local data processing github

    git clone https://github.com/PromtEngineer/localGPT.git

    22,193 stars · 3 of 39 lists · Python · MIT · pushed 2026-08-26

    Windows WSL2 macOS Linux Docker

    Context, Memory & RAG Local / Ollama

  7. 7 When one machine is not enough

    vLLM

    vllm-project/vllm

    High-throughput serving with paged attention. This is the step after a laptop -- still your hardware, but now several people can use it at once.

    A high-throughput and memory-efficient inference and serving engine for LLMs

    git clone https://github.com/vllm-project/vllm.git

    94,862 stars · 3 of 39 lists · Python · Apache-2.0 · pushed 2026-10-22

    Windows WSL2 macOS Linux Docker

    Harnesses & Runtime Infra Local / Ollama

The whole set, in one paste

Every install command above, in the order they are listed. Read them before you run them — they are other people's one-liners, and 2 of them pipe a download straight into a shell.

irm https://ollama.com/install.ps1 | iex
irm https://llama.app/install.ps1 | iex
git clone https://github.com/Mozilla-Ocho/llamafile.git
pip install open-webui
pip install gpt4all
git clone https://github.com/PromtEngineer/localGPT.git
git clone https://github.com/vllm-project/vllm.git