A team that uses AI models daily will sooner or later start paying for cloud chat: ChatGPT Plus for teams costs $21 per month, which amounts to $258 per year, and your conversations and uploaded documents remain on the provider's servers. Open WebUI is an open chat interface that runs in your browser like popular assistants, but you connect it to your own model sources: an Ollama server or any OpenAI-compatible API. Your chat history, team accounts, and uploaded documents stay on your Box, and you decide which provider you use and how much you pay. Below is a step-by-step guide on how to deploy Open WebUI on Saplo.
Why Open WebUI Instead of ChatGPT Plus
A paid AI chat in the cloud wins in simplicity, as long as only one person uses it independently. Problems start when the entire team needs access to the models:
- subscription charged per team seat, ChatGPT Plus for teams costs $21 per month, which amounts to $258 per year, and the bill grows with each additional team member who needs chat access,
- conversations and documents are stored by the provider, contracts, project descriptions, and internal notes uploaded to the chat are saved on servers of an external company, outside your company infrastructure,
- a single model provider, you can't connect your own model from Ollama’s server or a cheaper OpenAI-compatible API because the interface works only with the offer of one company.
Open WebUI doesn’t have any of these drawbacks: it’s a chat interface running on your Box, where you connect the model sources yourself. You can use the API key of any OpenAI-compatible provider, while at the same time keeping models on your own Ollama server and switch between them within the same conversation. You can create unlimited team accounts, and the total cost is just the price of the Box.
Requirements
To run Open WebUI on Saplo, you need:
- A Box with Docker running (the engine starts it automatically when installed from the app catalog),
- at least 1 GB of free RAM on the Box and 4 GB of disk space, so the cheapest plan that can handle this is Box S for $29 per year (chatting via API is light on the Box since computation happens on the model provider's side or your Ollama server),
- a domain or a *.saploapp.pl subdomain included with every plan, where the chat will be hosted,
- you do NOT need a separate database, Open WebUI stores data in its own SQLite database in a volume on your Box,
- access to models, meaning an API key from an OpenAI-compatible provider or the address of your own Ollama server. Saplo does not provide keys or its own Ollama; you connect the model source yourself after installation in the app panel.
You can find a full description of the app and its place in the catalog on the Open WebUI page on Saplo.
Step-by-step installation
- In the Saplo panel, go to the App catalog and select Open WebUI.
- Choose a Box with at least 1 GB of free RAM and 4 GB of free disk space, along with the domain or subdomain where the chat will be hosted.
- Confirm the installation. The engine downloads the official image
ghcr.io/open-webui/open-webui, sets up one container with a database in a volume, generates a persistent session secret (user accounts and login persist after container restarts), and exposes the panel through the Box's nginx with free SSL. - The first launch initializes the application database, including the vector database for searching uploaded documents, so it may take longer than the usual several seconds. The engine waits until the panel actually responds over HTTP before considering the installation complete.
- After the installation finishes, the chat is accessible at the specified domain. The first account you create in the registration form automatically receives administrator privileges.
Getting Started After Installation
- Open your app domain and create an account using the registration form. The first account you create will automatically become the administrator account, with no extra setup needed.
- Disable registration (Settings -> General -> Enable registration) so that people outside your team who visit your domain can't create accounts.
- Connect your model sources (Administration -> Settings -> Connections): paste your API key from an OpenAI-compatible provider or the address of your own Ollama server. You can connect multiple sources at once and switch models during a conversation.
- Add team accounts, and save shared instructions and prompts as ready-to-use models so everyone on the team starts with the same settings.
- Upload documents directly into conversations or to the shared knowledge base, then ask the model about their content without pasting entire files into the chat window.
Open WebUI on Saplo vs ChatGPT Plus
| ChatGPT Plus (team) | Open WebUI on Saplo | |
|---|---|---|
| Annual interface cost | $258 ($21/month) | $29 (Box S) |
| Fee per additional user | yes, subscription grows with team size | no, limited by Box resources |
| Models | single provider offer | any OpenAI-compatible API or your own Ollama |
| Changing model provider | new service, new history | just swap the key, history stays |
| Where conversations and documents are stored | on the provider's cloud | on your Box |
Besides that, the billing method matters: cloud subscription is paid continuously, even in months with no chat activity, while API costs fluctuate with your actual team usage.
Frequently Asked Questions
Does Saplo provide access to AI models? No, Open WebUI is an interface, not a model. You connect your own API key in the app panel (Administration -> Settings -> Connections). Any provider compatible with OpenAI works, and alternatively, you can connect your own Ollama server.
Do I pay for anything beyond the Box? If you use a model provider's API, that provider charges you based on usage, just like with any API key. With your own Ollama server, the cost depends on the hardware running the model. The Saplo interface itself is a fixed, known Box cost.
Do I need a separate database? No, Open WebUI saves everything in its own SQLite database in a data volume on your Box. The Saplo engine sets up one container, with no need for MySQL or Postgres to buy or configure.
How many people can use the chat? Open WebUI does not count users: you can create unlimited accounts without extra fees for additional accounts. The real limit is the Box's resources, and chatting via the API only lightly uses Box, as the counting happens on the model provider’s side.
Are my conversations private? Chat history, accounts, and uploaded documents are stored in the database on your Box. The question content goes only to the provider whose key you connected, so you decide who processes your team's data.
Open WebUI doesn’t have to run alone: the same Box can also run other apps from the app catalog, within its resource limits. You can find details on the app hosting page.
If your team is paying for AI chat in the cloud but wants its own interface, free choice of models, and conversations on your own infrastructure, write to us. We’ll help you pick a Box and set up Open WebUI.
- #AI
- #Ollama
- #Open WebUI
- #Guide
