Guide · Open source
How to use GPT4All: install, chat locally and use Python
Updated · written with AI from the project's own documentation and checked against our catalogue data
GPT4All is an open-source tool for running large language models privately on everyday desktops and laptops. This guide shows you how to download the application, check whether your computer meets the requirements, chat with a local model, and use the Python client if you want to call models from code.
GPT4Allnomic-ai/gpt4all · ★ 77.3k · MIT Install options & downloads Quick start
- 1Check your hardware. Confirm your processor and operating system meet the minimum requirements listed in the README.
- 2Download the installer. Choose the installer for your platform from the download links on the project page.
- 3Open the application. Run the installer and launch GPT4All from your applications list.
- 4Pick and download a model. Select a model in the app and wait for it to finish downloading to your computer.
- 5Start chatting. Type a question in the chat box and read the local model's reply.
- 6Optional: install the Python client. Run pip install gpt4all and load a model file in your code if you want to use it programmatically.
Check whether your computer meets the requirements
GPT4All runs models on your own machine, so your hardware sets the limits. The README lists the minimum processor for Windows and Linux as Intel Core i3 2nd Gen or AMD Bulldozer, or better.
The Windows ARM build supports Qualcomm Snapdragon and Microsoft SQ1/SQ2 processors. The Linux build is x86-64 only and does not support ARM. The macOS build requires Monterey 12.6 or newer, and the README says the best results come on Apple Silicon M-series processors.
- Windows and Linux: Intel Core i3 2nd Gen / AMD Bulldozer or better
- Windows ARM: Qualcomm Snapdragon or Microsoft SQ1/SQ2
- Linux: x86-64 only, no ARM
- macOS: Monterey 12.6 or newer, best on Apple Silicon
Download and install the GPT4All application
The README links a separate installer for each platform: Windows (x64 and ARM), macOS and Ubuntu. Download the one that matches your system and run it like any other desktop application. The README says you can start without API calls or GPUs.
GPT4All is also listed on Flathub, which the README describes as community maintained. The full System Requirements page in the project's repository has more detail if the minimum specifications are unclear.
- Windows: gpt4all-installer-win64.exe (or the ARM installer)
- macOS: gpt4all-installer-darwin.dmg
- Ubuntu: gpt4all-installer-linux.run
- Flathub: io.gpt4all.gpt4all (community maintained)
Chat with a local model in the desktop app
Once the app is open, you can chat with a language model on your machine. Because the model runs locally, the text you type is not sent to a cloud service, according to the project's description of how it works.
The desktop app includes LocalDocs, which lets you privately and locally chat with your own documents. The release history describes the V3.0.0 release, from July 2024, as including an improved user workflow for LocalDocs.
- Pick a model from the app's model list and wait for it to download
- Type your question in the chat box
- Use LocalDocs to ask questions about your own files
Use GPT4All from Python
The Python client, gpt4all, gives you access to local models through llama.cpp implementations. Install it with pip, then load a model file by name. The README example downloads or loads a 4.66GB model, so make sure you have the disk space first.
The example uses a chat session and the generate method with a token limit. The model file named in the example is a GGUF file.
pip install gpt4all
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))Connect GPT4All to other tools
The README lists integrations with LangChain, the Weaviate vector database and OpenLIT for monitoring. Each has its own documentation linked from the project README.
The release history also mentions a Docker-based API server that exposes local models through an OpenAI-compatible HTTP endpoint. Check the linked project page for setup details before relying on it.
- LangChain
- Weaviate Vector Database
- OpenLIT (OTel-native monitoring)
Know the limits before you commit
Local models are limited by your machine's memory, processor and graphics hardware. The README notes that they do not match the most powerful cloud-hosted models. Very old computers and ARM-based Linux machines are not supported by the build described in the README.
The README also notes that GPT4All welcomes contributions. Its documentation, Discord and issue tracker are linked from the project page if you get stuck.
What people use it for
- Chatting with a language model without sending your text to a cloud service
- Asking questions about your own documents with LocalDocs
- Calling a local model from Python scripts with the gpt4all client
- Trying local AI on an ordinary laptop without a dedicated graphics card
Frequently asked questions
Is GPT4All free to use?
GPT4All is open source under the MIT licence, and the README says it is available for commercial use. Check the licence in the repository if you plan to redistribute it.
Does GPT4All need a GPU?
The README says no GPUs are required and you can just download the application to get started. Performance will depend on your processor and memory.
What are the GPT4All system requirements?
The README lists Intel Core i3 2nd Gen or AMD Bulldozer or better for Windows and Linux, the Linux build is x86-64 only, and macOS needs Monterey 12.6 or newer. The full System Requirements page has more detail.
Does GPT4All send my data to the cloud?
The project describes GPT4All as running models privately on your own device with no API calls required. Your chats are processed locally by the model you load.
How does GPT4All compare to Ollama or LM Studio?
The comparable projects listed are llama.cpp, Ollama, LM Studio and Jan. GPT4All is a desktop app with a Python client and LocalDocs, so choose based on which interface and features you need.
Why won't GPT4All install on my older computer or ARM Linux machine?
The README lists minimum processor requirements for Windows and Linux, and says the Linux build is x86-64 only. Check your processor against the requirements before downloading.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
GPT4All suits people who want private, local chat on an ordinary laptop, and developers who want local models in Python. It is a poor choice if you need the most powerful cloud-hosted models, or if you run very old hardware or ARM Linux.
Install options & downloads