REVIEW · LOCAL CHAT APP
Verdict: Yes. GPT4All still offers a macOS download, runs AI models privately on your Mac, and LocalDocs lets those models answer from folders you choose, citing the files it used. It is a chat tool that reads collections you set up, not a search engine for finding one lost file.
11 November 2026 · By Ishtiaque Hossain · 4 minute read
Looking for a file you can describe but not name? Try Shlexball. The first 30 searches are free and need no account.
macOS 15 or later. Apple silicon. 16 GB memory. On a phone? Email yourself the link.
This review is based on GPT4All's own pages, read on 14 October 2026. Every claim links to the page it came from. We are the makers of Shlexball, so we say plainly where GPT4All does better.
| Price | Free and open source (MIT licence) |
| Runs on | macOS Monterey 12.6 or newer, best on Apple silicon; also Windows and Linux |
| LocalDocs | Index a folder and let the AI answer from it, with sources shown |
| Where your data goes | Runs on your device; no cloud required |
| Made by | Nomic AI |
| Best for | Private AI chat over a set of your own documents |
GPT4All is a private, local AI chatbot from Nomic. It runs open-source language models on everyday desktops and laptops, with no API calls or graphics card required.
Nomic says no cloud is required and no data leaves your machine, and it offers access to thousands of models with full customization.
The project is open source under the MIT licence, popular on GitHub with about 77.4k stars, and also offers a Python client for developers.
LocalDocs brings information from files on your device into your chats, privately. You click Add Collection, name it and link it to a folder.
GPT4All then indexes that folder with Nomic's free on-device embedding models. Each text snippet gets an embedding vector, a list of numbers that represents its meaning.
When you ask a question, LocalDocs finds snippets that are semantically similar, meaning close in meaning, and adds them to the prompt so the model can answer from your files.
Progress shows on the LocalDocs page and a green Ready indicator appears when a collection is done; you can chat with the files already processed before that. Click Sources under an answer to see which files it used.
Keep each collection focused on one folder of related files, so the snippets LocalDocs finds are relevant to the questions you ask in that chat.
Open LocalDocs with the button in the top-right corner of a chat to give the model context from your collections, and check Sources below each answer to see which files it drew on.
If you change your LocalDocs settings later, you can rebuild your collections with the new settings.
Developers can go further: the gpt4all Python client wraps llama.cpp, the project integrates with Langchain and Weaviate, and Nomic recommends its Python SDK for trying the embedding models directly.
Yes. Nomic offers a macOS download, and the macOS build requires Monterey 12.6 or newer, with the best results on Apple silicon M-series processors.
Its README highlights support for DeepSeek R1 distilled models. The release history in that README lists version 3.0.0, from July 2024, as its latest entry, which brought a redesigned chat interface and an improved LocalDocs workflow.
Nomic's wider focus has also grown: its site now leads with the Nomic Platform, AI agents for architecture, engineering and construction workflows.
Privacy is GPT4All's core promise: AI on your device, no cloud required and no data leaving your machine.
LocalDocs follows the same approach, using on-device embedding models to index your folders.
Nomic designed it for developers, teams and AI power users who want control over models and data.
If you want to ask questions about a folder of documents without sending them anywhere, LocalDocs does that well. If you want to find which file holds something across your whole Mac, you will be setting up collections for every folder first.
Shlexball finds a file on your Mac from a plain-English description of what you remember, searching file and folder names and the text inside files, including PDFs and Word documents, without collections to set up.
Like GPT4All, it runs a language model on your Mac and never uploads files. The first 30 searches are free.
Built to find files, with ranked results and previews: Shlexball, 30 free searches.
▶ DOWNLOAD SHLEXBALLA feature that indexes a folder you choose with on-device embedding models, then adds the most relevant snippets to your chat so the AI can answer from your files.
Yes. The macOS build requires Monterey 12.6 or newer and works best on Apple silicon M-series processors.
Yes. It is open source under the MIT licence.
Nomic says GPT4All runs on your device with no cloud required and no data leaving your machine.
Click Sources below the response to see which files from your LocalDocs collection were referenced.