Deploy Open WebUI on Northflank

Open WebUI is an open-source interface for working with large language models through a clean and familiar chat experience. It supports multiple model providers, including Ollama and any OpenAI-compatible API, and is widely used for private ChatGPT-style deployments.

With Northflank, you can deploy Open WebUI in minutes using this stack template. The template handles storage, networking, and environment configuration, and allows you to connect the UI to an existing Ollama deployment.

If you don’t already have one, you can deploy our Ollama GPU stack template to spin up a fully managed Ollama instance with GPU acceleration and then link it to Open WebUI using OLLAMA_BASE_URL.

Open WebUI is a self-hosted chat interface that brings together model interaction, document retrieval, browsing, and multimodal capabilities in a single environment. It is designed for teams and developers who want a private conversational AI workspace without relying on third-party platforms.

You can connect Open WebUI to Ollama, import documents for retrieval-augmented conversations, enable web browsing, use voice input, or add image generation through OpenAI, Stable Diffusion, or other compatible APIs. The platform is fully configurable and works well for personal use or internal teams that need a secure and extensible AI chat environment.

The Open WebUI stack template provisions everything required to run a full AI chat interface with persistent storage and a connection to your Ollama service.

It includes:

• A deployment service running the official ghcr.io/open-webui/open-webui:main image
• A persistent volume mounted at /app/backend/data to store settings, chats, uploads, and configuration
• A secret group for securely storing the required OLLAMA_BASE_URL for model inference

This configuration ensures that your chat environment remains stable, persists data correctly, and integrates smoothly with your existing model runtime.

1. Create an account on Northflank
2. Click deploy Open WebUI now to open the stack template
3. Provide the OLLAMA_BASE_URL from your Ollama deployment
4. Click deploy stack to create the project, secrets, service, and storage
5. Wait for the deployment to complete
6. Open the service and visit the public URL to access your WebUI instance

Once the service is live, you can sign in, configure your workspace, and begin interacting with your connected models.

If you are using Ollama, you can pull a model and begin chatting immediately. For example:

curl https://YOUR_OLLAMA_URL/api/pull -d '{
  "name": "qwen2.5"
}'

After the model is installed, Open WebUI will be able to use it for chatting, agents, document retrieval, and any configured workflows.

This template provides a complete private AI chat environment with the following benefits:

• A browser-based interface for interacting with LLMs
• Persistent storage for chats, user data, settings, and documents
• Full support for Ollama models and OpenAI-compatible APIs
• Document imports for retrieval-augmented conversations
• Optional browsing, voice input, and image generation integrations
• Secure environment variables for configuring backends and keys
• Automatic HTTPS and public access through a single service URL

It follows a clean and production-ready structure that works for both personal use and internal deployments.

• WebUI Service - Runs the Open WebUI backend and frontend, manages authentication, chat sessions, and model requests.
• Persistent Volume - Stores all user data, uploaded documents, conversations, settings, and internal configuration files.
• Secret Group - Contains the OLLAMA_BASE_URL used by the service to route generation requests.

Deploying Open WebUI on Northflank gives you a secure, private, and fully featured interface for working with large language models. With persistent storage and built-in support for Ollama, you can run a complete ChatGPT-like environment without managing servers or infrastructure.

You now have everything you need to build a customizable AI workspace for yourself or your team, backed by a stable and production-ready deployment.