# How to Deploy AnythingLLM With xCloud One Click Apps

> Deploy AnythingLLM on xCloud from One Click Apps: pick a server with 2 GB RAM or more, set an instance password, choose a domain, then connect your LLM provider.

You can run [AnythingLLM](https://anythingllm.com/) on your own server with xCloud without writing a Docker Compose file. Click **New Site**, pick **AnythingLLM** in the app catalog, choose a server, enter a site title, choose **Demo Site** or **Go Live**, and click **Next**. xCloud starts AnythingLLM with a password, keeps its data on the server and serves it over HTTPS. Then you sign in and connect the AI provider you want to use. This guide walks through each screen.

**Who this is for:** teams who want a private AI workspace to chat with their own documents, build knowledge bases and run AI agents, on a server they control. For server sizes and prices, see [AnythingLLM hosting on xCloud](/anythingllm-hosting/).

## What the AnythingLLM app includes

| Part | What you get |
|---|---|
| AnythingLLM | AnythingLLM 1.16.2, from the official AnythingLLM image, in one container. |
| Search over your documents | AnythingLLM searches your uploaded documents by itself, so you don't need any extra service. |
| AI model | Not included. You connect a provider (OpenAI, Anthropic, Azure, an Ollama server you run elsewhere, any OpenAI-compatible service and more) after the install. |
| Web address | AnythingLLM is only reachable through your domain, with an SSL certificate, not by server IP and port. |
| Password | The whole instance is protected by a password from the first start. |
| Saved data | Workspaces, chats, users, uploaded documents, search indexes and the settings you save in AnythingLLM, including provider API keys, are kept on the server, so they survive restarts and redeploys. |

AnythingLLM is open source under the MIT licence.

## Before you start

- An xCloud account that can create sites.
- An **Nginx** server with Docker and at least **2 GB RAM, 2 CPU cores and 10 GB disk**. You can also create a new server during setup.
- An API key from the AI provider you want to use. Its usage is billed to your provider account.
- For **Go Live**: a domain you can point at the server. For a first try, **Demo Site** needs no domain.

## Step 1: Find AnythingLLM in the app catalog

From the xCloud dashboard, click **New Site**. On **What would you like to host today?**, type **AnythingLLM** in the search box, then click the **AnythingLLM** card.

![The AnythingLLM card in the xCloud app catalog after searching for AnythingLLM](/_landing/docs/deploy-anythingllm-with-xcloud-image-1.png)

The AnythingLLM page opens. It shows **Runs on Nginx, Docker** and the recommended size: **2 GB RAM · 2 vCPU · 10 GB disk**.

## Step 2: Choose where AnythingLLM will run

Under **Where should it run?**, pick one option:

- **Deploy on a new server:** xCloud creates and sets up a server only for AnythingLLM.
- **An existing server:** click its card. Each card shows the server's RAM, CPU and free disk.

Click **Continue**.

![An existing Nginx server selected under Where should it run? on the AnythingLLM install page](/_landing/docs/deploy-anythingllm-with-xcloud-image-2.png)

A server with less than 2 GB of RAM, or on the wrong stack, is listed under **server cannot run this app**. Open that list to see why.

## Step 3: Add the site title, domain and password, then install

The **Install AnythingLLM Into** form opens.

1. Enter a **Site Title**, for example `AnythingLLM`. **Add Tag (optional)** helps you group sites.
2. Under **Domain Setup**, choose one:
   - **Demo Site:** xCloud gives you a test domain, shown under **Domain Name**. Click **Regenerate** for a different one. Use this to try AnythingLLM first.
   - **Go Live:** enter your own domain and point its DNS at the server.
3. Under **Access**, enter an **Instance Password**, or leave it empty and xCloud creates a strong one. Use 12 to 64 characters: letters, numbers, dashes and underscores only.
4. Click **Next**.

![The Instance Password field under Access in the Install AnythingLLM Into form](/_landing/docs/deploy-anythingllm-with-xcloud-image-3.png)

## Step 4: Wait for the setup to finish

xCloud shows a progress page with each stage: checking the server and DNS, installing Docker, setting up the app, SSL, HTTPS, Nginx and monitoring. You can leave this page; the setup keeps running.

![The Setting Up Your Site progress page while xCloud installs AnythingLLM](/_landing/docs/deploy-anythingllm-with-xcloud-image-4.png)

When it finishes, the site's **Site Overview** page opens.

If a stage fails, open **Recent Events** on the site, click **View** on the failed event, and fix what it reports (often DNS for **Go Live**). Then try again, or contact support with the event output.

## Step 5: Copy your AnythingLLM address and password

On **Site Overview**, the **AnythingLLM Credentials** card shows the **URL** and the **Password**. Click **Copy** next to each one. Click the eye icon to show the password.

![The AnythingLLM Credentials card with the URL and Password on Site Overview](/_landing/docs/deploy-anythingllm-with-xcloud-image-5.png)

## Step 6: Sign in and connect your AI provider

Open the **URL**. AnythingLLM shows **Welcome** with a **Password** field. Paste the password and click **Login**.

![The AnythingLLM Welcome screen with the Password field and the Login button](/_landing/docs/deploy-anythingllm-with-xcloud-image-6.png)

On its first start AnythingLLM can take a minute or two to be ready. If the page doesn't load yet, wait a minute and reload.

AnythingLLM then opens its setup wizard:

1. Choose your LLM provider and enter its API key.
2. Create your first workspace.
3. Upload documents or add web pages to the workspace, then start a chat.

The [AnythingLLM documentation](https://docs.anythingllm.com/) covers providers, workspaces, agents and the API.

## After the install

### Change the password

The instance password lives in the site's settings in xCloud, and AnythingLLM applies it on every start. To change it:

1. Open the site in xCloud and click **Environment** in the site menu.
2. Change `AUTH_TOKEN` to the new password.
3. Click **Save & Redeploy**.

A password changed inside AnythingLLM works only until the next restart, then goes back to the value in **Environment**. For the same reason, if you turn off password protection under **Settings → Security** in AnythingLLM, it comes back on at the next restart.

Never change `SIG_KEY` or `SIG_SALT` in **Environment**. AnythingLLM uses them to protect sign-in sessions and some saved secrets.

### Give each person their own login

By default everyone shares one password. To give each person their own account, open **Settings → Security** in AnythingLLM, turn on **Multi-User Mode** and create an admin user. This replaces the shared password with personal logins, and you can't switch it back.

### Keep usage under control

- Anyone who can sign in can use your AI provider, and the cost goes to your provider account. Share the password, or create accounts, only for people who need it.
- Agent skills, MCP servers and web browsing let the AI reach other systems. Turn them on only when you need them.

### Upload size

AnythingLLM accepts one document up to **512 MB** through your domain. If your domain goes through Cloudflare's proxy, Cloudflare limits one upload to about **100 MB**.

### Server size

The minimum of 2 GB RAM suits a few users and a modest document library. Large document collections and many users at once need more. Running an AI model on the same server needs far more RAM; see [Self-host LLMs with xCloud](/docs/how-to-self-host-llms-with-xcloud/) for that setup.

## Update and back up AnythingLLM

- **Update:** your site stays on the AnythingLLM version it was installed with. Back up first, read the [AnythingLLM release notes](https://github.com/Mintplex-Labs/anything-llm/releases), then follow [Update a One Click App and edit its Docker Compose](/docs/update-a-one-click-app-and-edit-its-docker-compose-in-xcloud/). AnythingLLM updates its own database when it starts. Never remove the data volume during an update.
- **Back up:** one data volume holds everything: workspaces, chats, users, documents, search indexes and the saved settings with your provider API keys. Keep backups private. See [Back up and restore Docker apps](/docs/backup-and-restore-docker-apps/) for what xCloud backs up and how to restore it.

## Common mistakes

- Expecting a built-in AI model. Connect a provider and enter its API key after you sign in.
- Changing the password inside AnythingLLM. It goes back at the next restart; change `AUTH_TOKEN` in **Environment** instead.
- Turning on **Multi-User Mode** to test it. It can't be turned off again.
- Choosing **Go Live** before the domain's DNS points at the server. Start with **Demo Site**, then [go live from the demo site](/docs/how-to-go-live-from-demo-site/).

## Frequently asked questions

### Do I need to write a Docker Compose file to deploy AnythingLLM on xCloud?

No. The AnythingLLM One Click App already defines the AnythingLLM container and its storage. You pick AnythingLLM in the app catalog, choose a server, enter a site title, choose Demo Site or Go Live, and click Next.

### What server does AnythingLLM need on xCloud?

An Nginx server with Docker and at least 2 GB of RAM, 2 CPU cores and 10 GB of disk. Large document collections and many users need more.

### Does the AnythingLLM app include an AI model?

No. It includes AnythingLLM with built-in document search, but no language model. After the install you connect a provider such as OpenAI, Anthropic, Azure or an Ollama server you run elsewhere, and enter its API key in AnythingLLM.

### How do I change the AnythingLLM password?

Open the site in xCloud, click Environment, change AUTH_TOKEN and click Save & Redeploy. A password changed inside AnythingLLM goes back to the Environment value at the next restart.

## Related guides

- [Deploy LibreChat with xCloud](/docs/how-to-deploy-librechat-with-xcloudone-click-apps/)
- [Use One Click App templates in xCloud](/docs/how-to-use-one-click-app-templates-in-xcloud/)
- [Back up and restore Docker apps](/docs/backup-and-restore-docker-apps/)
- [Go live from a demo site](/docs/how-to-go-live-from-demo-site/)

If you run into any issues while deploying AnythingLLM, feel free to reach out to our [support team](/docs/access-built-in-support-portal-in-xcloud/).