# GPT4All

> Run large language models privately on everyday laptops and desktops

- Category: Models & Inference / Local Runtimes
- Language: C++
- Licence: MIT
- Repository: https://github.com/nomic-ai/gpt4all
- Website: https://nomic.ai/gpt4all
- Canonical page: https://ai-tldr.dev/tools/gpt4all/

## In short

GPT4All, from Nomic AI, lets you run large language models (LLMs) directly on everyday desktops and laptops. You download the application, pick a model, and chat with it locally. It sits in the Local Runtimes category of the Models & Inference tools and is open-source, written in C++ and licensed MIT.

## Overview

GPT4All, from Nomic AI, lets you run large language models (LLMs) directly on everyday desktops and laptops. You download the application, pick a model, and chat with it locally. No API calls or GPUs are required, so your conversations stay on your own machine.

It ships as a desktop chat app for Windows, macOS, and Ubuntu, plus a Python client built around llama.cpp implementations. The app includes LocalDocs, a feature that lets you privately chat with your own files, and it can run a range of open model architectures.

## What it does

- Desktop chat app for Windows, Windows ARM, macOS, and Ubuntu
- Runs models fully locally with no API calls or GPU needed
- Python client that wraps llama.cpp for programmatic use
- LocalDocs lets you chat privately with your own documents
- Nomic Vulkan support for local GPU inference on NVIDIA and AMD cards
- OpenAI-compatible HTTP endpoint for serving local models

## Getting started

The fastest way to try GPT4All is the desktop installer, but you can also use the Python client to run models from code.

1. Install the Python client

   Install the gpt4all package from PyPI to access LLMs from Python.

```bash
pip install gpt4all
```

2. Load a model and chat

   Create a GPT4All object with a model file. The first run downloads the model, then you can generate text inside a chat session.

```python
from gpt4all import GPT4All
model = GPT4All("Meta-Llama-3-8B-Instruct.Q4_0.gguf") # downloads / loads a 4.66GB LLM
with model.chat_session():
    print(model.generate("How can I run LLMs efficiently on my laptop?", max_tokens=1024))
```

3. Or download the desktop app

   Prefer a graphical chat window? Grab the installer for your platform from gpt4all.io (Windows, Windows ARM, macOS, or Ubuntu) and follow the quickstart guide in the documentation.

## When to use it

- Chatting with an AI assistant offline, keeping all data on your own computer
- Asking questions about your private files and notes with LocalDocs
- Building local AI features in Python without sending data to a cloud API
- Serving a local model through an OpenAI-compatible endpoint for apps and tools