DeepTutor is a free, open source study app from the HKUDS research group at the University of Hong Kong. You give it your notes, PDFs and slides, and it becomes a personal tutor: it answers questions from your own material, makes quizzes, plans what to study next and turns a topic into a book or a short video lesson. It installs like a normal app on Mac and Windows, it has no paid plan, and it can run on a free AI model on your own computer through Ollama. It was one of the fastest-growing projects on GitHub in November 2026, passing 58,000 stars less than a year after its first release. It's Apache 2.0 licensed, so you can use it, change it and build it into a product you sell.
About DeepTutor
DeepTutor is a free AI tutor that you install on your own computer. Instead of pasting notes into ChatGPT one chat at a time, you put your study material into DeepTutor once, and it remembers what you're learning, what you got wrong and what to cover next. It was made by the HKUDS research group at the University of Hong Kong, first released at the end of December 2025, and it updates almost every week.
What it can do
Answers from your own material. Upload PDFs, Word and PowerPoint files, Markdown or plain text into a knowledge base, and DeepTutor answers with your sources in mind instead of guessing.
Quizzes and a question bank. Ask for a quiz on any topic or chapter. Questions you get wrong are saved so you can practise them again.
Solve, step by step. For maths, science and coding problems, Solve works through the answer and explains each step rather than just giving the result.
Deep research. Ask a bigger question and it searches, reads and writes you a report with sources.
Books and video lessons. Turn a topic, or your own notes, into a "living book" you can read and add to, or a short video lesson.
Mastery Path. Tell it what you want to learn and it plans the steps, tracks how you're doing and adjusts.
Memory you can see. It keeps notes on how you learn and what you already know, and you can read and edit that memory yourself.
Your choice of AI. Use a free model on your own computer through Ollama or LM Studio, sign in with a Claude or ChatGPT plan you already have, or add an API key from OpenAI, Anthropic, DeepSeek, Gemini and many others.
Where you'll meet it
DeepTutor doesn't come with its own AI model. It sits on top of one, so it pairs well with Ollama or llama.cpp when you want everything free and private on your own computer. You can also run it in your web browser with Docker, add your own study skills, and connect it to chat apps so your tutor can message you.
DeepTutor, NotebookLM or Open Notebook?
All three let an AI read your documents and help you understand them. They differ in how much they do for studying, where your files go and what they cost.
DeepTutorThis page
Best forA full personal tutor for studying
Quizzes, step-by-step solving, study plans, books and video lessons from your own material, with memory of how you learn. Can run fully offline on a local model.
How you use it
A desktop app for Mac and Windows, or in your browser with Docker
Open source
Yes (Apache 2.0)
Cost
Free; you bring the AI model
NNotebookLM
Best forQuick summaries and audio overviews
Google's polished notebook for asking questions about your documents, with podcast-style audio summaries. Nothing to install, but your files are uploaded to Google.
How you use it
In your web browser
Open source
No
Cost
Free; higher limits with a paid Google AI plan
OOpen Notebook
Best forA private NotebookLM you host yourself
An open source take on NotebookLM, with notes, sources and podcast-style summaries, and support for local models through Ollama. Built for research rather than tutoring.
How you use it
In your web browser, after installing it with Docker
Open source
Yes (MIT)
Cost
Free; you bring the AI model
A quick way to choose: want summaries of a few documents right now, with nothing to install? NotebookLM. Want a private research notebook on your own computer? Open Notebook. Studying for something and want quizzes, worked solutions and a plan? DeepTutor.
Good to know
The desktop app is still marked "Preview". On Windows it isn't signed yet, so SmartScreen may show a warning the first time you open it.
DeepTutor is free, but the AI behind it isn't always. A model on your own computer costs nothing; an online model is billed by the company that runs it, usually a few dollars a month for normal studying.
Small models on a laptop are fine for chat and quizzes. Research and Solve work much better with a bigger local model or an online one.
To build a knowledge base you also need an embedding model, which turns your files into something the AI can search. Ollama's free nomic-embed-text works well.
If you change the embedding model later, you have to rebuild your knowledge bases.
Licence: Apache 2.0. Free to use, change and include in products you sell, as long as you keep the licence and copyright notices. The AI models you use with it have their own licences.
What you can build with it
Studying for exams from your own notes, slides and PDFs
Making practice quizzes and flashcard-style question banks
Getting step-by-step help with maths, science and coding problems
Learning a new skill with a study plan that adapts to you
Running a private AI tutor on your own computer with no subscription
Building blocks
DeepTutor is made of parts you can use on their own in your project. Copy a prompt to have your AI assistant add one for you, and see the licence card for what you may do with them.
Desktop app for Mac and Windows
Free to use, even in closed products
The full tutor as a normal app you install, with everything it needs built in. Download it, open it and pick an AI model in Settings. Still marked Preview.
Download the .dmg (Mac) or Setup.exe (Windows) from github.com/HKUDS/DeepTutor/releases/latest
Licence
Apache 2.0 (read from LICENSE (Apache License 2.0) at the repository root)
Run it in your browser with Docker
Free to use, even in closed products
The official Docker image runs the whole web app at http://127.0.0.1:3782, with your settings and notes kept in a Docker volume. Good for a home server or a shared study computer.
docker run --rm --name deeptutor -p 127.0.0.1:3782:3782 -v deeptutor-data:/app/data ghcr.io/hkuds/deeptutor:latest
Licence
Apache 2.0 (read from LICENSE (Apache License 2.0) at the repository root)
Python package and command line
Free to use, even in closed products
Install DeepTutor with pip, set it up with deeptutor init and start it with deeptutor start. Also lets you build knowledge bases and quizzes from the terminal or your own scripts.