Dify: How to Self-Host It With Docker Compose on Windows and Mac

By Shah Rukh, software developer · October 2, 2026 · 8 min read

Dify is an open-source platform for building AI apps, such as chatbots, document assistants and multi-step workflows, in a visual editor in your browser. This guide shows how to run the free Community Edition on your own Windows PC or Mac with Docker Compose, what it really costs, and what the licence does and does not allow.

Read this first. Dify is free to self-host, but it does not include an AI model. You connect your own model provider, and most cloud providers charge for API use. Dify also runs as a group of more than ten Docker containers, so it is heavier than a normal desktop app. And its licence is not plain Apache 2.0: it has extra conditions, explained below.

What Dify is

The project describes Dify as an “open-source LLM app development platform”. An LLM (large language model) is the kind of AI behind chat assistants. Dify gives you a dashboard where you design what the AI should do, instead of writing all the code yourself.

The README lists these main features:

• Workflow: build and test AI workflows on a visual canvas.
• Model support: the project says it works with hundreds of models from dozens of providers, including any OpenAI API-compatible model and self-hosted ones.
• Prompt IDE: write prompts and compare how different models answer.
• RAG pipeline: RAG means the AI answers from your own documents. Dify can extract text from PDFs, PPTs and other common formats.
• Agent: agents that run commands, install software and handle files inside their own sandbox (an isolated container).
• LLMOps: logs and performance monitoring for your apps.
• Backend-as-a-Service: everything you build also gets an API, so your own website or program can call it.

What you need

• CPU: at least 2 cores. RAM: at least 4 GiB. These are the minimums in the README; more memory makes it far more comfortable.
• Docker and Docker Compose v2.24.0 or later. Docker runs software in isolated boxes called containers. Compose starts many containers together.
• Git, or a ZIP download of the repository from GitHub.
• A good internet connection for the first start. Docker downloads images for Dify’s API and web app, PostgreSQL, Redis, Weaviate, Nginx, a sandbox, a plugin daemon and more. The repository does not state the total download size.
• An API key from a model provider, or a model running on your own machine. This is the part that can cost money.

No GPU is needed for Dify itself. You only need a strong machine if you also run the AI model locally.

What the Dify licence allows

The repository uses the “Dify Open Source License”, which is Apache 2.0 with additional conditions. In plain words:

• Commercial use is allowed, including using Dify as the backend of your own application or as an internal platform in a company.
• No multi-tenant service without permission. You may not use the Dify source code to run a multi-tenant environment unless Dify authorises it in writing. The licence defines one tenant as one workspace. If you want to do this, you need a commercial licence.
• Keep the logo and copyright. If you use Dify’s frontend (the web/ folder or the “web” Docker image), you may not remove or change the logo or copyright information in the console or apps. This does not apply if you don’t use the frontend.
• Contributors agree the company can make the licence stricter or looser later, and can use contributed code commercially.
• The licence also notes that the product’s interactive design is protected by an appearance patent.

For personal use on your own computer, none of these get in your way. If you plan to resell Dify to clients as a hosted service, read the LICENSE file yourself and ask the company. This is a summary, not legal advice.

Precautions before you install

1. Use the real repository. Clone only from github.com/langgenius/dify.
2. The sample passwords are public. .env.example ships with DB_PASSWORD=difyai123456, REDIS_PASSWORD=difyai123456, a fixed WEAVIATE_API_KEY, PLUGIN_DAEMON_KEY and SANDBOX_API_KEY=dify-sandbox. The file itself marks DIFY_AGENT_API_TOKEN and DIFY_AGENT_SERVER_SECRET_KEY as development defaults to replace in production. Fine for a test on your laptop; change them before anyone else can reach the machine. Note the Redis password also appears inside CELERY_BROKER_URL, so change both together.
3. It listens on port 80. The Nginx container publishes ports 80 and 443 on your computer, and the plugin debugging port 5003 is published too. On shared Wi-Fi, other devices may be able to open your Dify login page. Keep your firewall on and don’t forward these ports from your router.
4. Create the admin account straight away. The first person to open the install page sets up the admin account.
5. Your data goes to the model provider. Prompts and the document text used to answer them are sent to whichever provider you connect. Don’t upload client or private files unless that provider’s terms suit you, or use a local model.
6. It contacts Dify’s servers. The defaults include CHECK_UPDATE_URL=https://updates.dify.ai and MARKETPLACE_ENABLED=true (the plugin marketplace). Install only plugins you trust.
7. Your secrets live in two places. Settings are in docker/.env. Uploaded files and databases are in docker/volumes/. Don’t share or publish either.
8. Agents and code nodes run code. They run inside sandbox containers, and SANDBOX_ENABLE_NETWORK=true by default, so they can reach the internet through a proxy. Review what you ask an agent to do.

How to install Dify on Windows

The repository gives one set of commands and no Windows-specific steps. They work in PowerShell once Docker is installed.

1. Install Docker from docs.docker.com/get-docker (the link the README uses), start it and wait until it says it is running.
2. Open PowerShell and download the project:

   git clone https://github.com/langgenius/dify.git

3. Run the README’s four commands, one line at a time:

   cd dify
   cd docker
   cp .env.example .env
   docker compose up -d

4. Wait. The first run downloads every image and can take a long time on a slow connection.

In PowerShell, cp is a built-in shortcut for copying a file. In the old Command Prompt it does not exist, so use PowerShell.

How to install Dify on Mac

1. Install Docker from docs.docker.com/get-docker and start it.
2. Open Terminal (Cmd + Space, type Terminal).
3. Run:

   git clone https://github.com/langgenius/dify.git
   cd dify
   cd docker
   cp .env.example .env
   docker compose up -d

The same commands apply on Linux. The repository does not say anything specific about Apple Silicon or Intel Macs.

First use: create your admin account

1. Open http://localhost/install in your browser. This starts what the README calls the initialization process.
2. Create the admin account with a strong, unique password.
3. Log in and add a model provider in the settings. Paste your API key there. The README links a full list of supported providers in the official documentation.
4. Create your first app and send a test message.

The exact screens are not described in the repository, so follow what you see on the page.

Useful settings in the .env file

Edit docker/.env, then run docker compose up -d again from the docker folder to apply changes.

Optional advanced settings live in the docker/envs/ folder. Copy one of its .env.example files next to itself without the .example ending to use it. Values in docker/.env win over those files.

How to update Dify, and what about uninstalling

The repository does not give a full update command list. It does provide a helper for your settings file. After pulling a newer version, new settings may appear in .env.example. This script adds them to your .env, keeps your own values, and first saves a backup in env-backup/:

chmod +x dify-env-sync.sh
./dify-env-sync.sh

It is a shell script, so it suits Mac and Linux. There is also a dify-env-sync.py file in the same folder, but the README does not explain how to run it. Copy the docker/volumes/ folder somewhere safe before any update.

The project does not document an uninstall. You can stop and delete the containers from the Docker app’s own window. Your data stays in docker/volumes/ until you delete that folder.

Common problems

For anything else, the README points to the official FAQ and to GitHub Discussions.

This guide is based on the project’s repository as it was on October 2, 2026, when the Docker files used Dify version 1.17.1. The project changes often, so if a step looks different, follow the README on GitHub. ToolsCloset is not connected to the project or to LangGenius, Inc.

More AI tips

Related tools