Open WebUI Hosting

How Open WebUI works

Open WebUI is the conversation layer between users and one or more AI providers. Administrators add a connection to a remote OpenAI-compatible endpoint, a remote Ollama service, Anthropic, or another supported provider, and the interface exposes the available models to authorised users. Prompts are sent to the selected provider, which performs inference and returns the response to the chat workspace.

The application keeps conversations and user-facing settings in the hosted workspace, while the model may run somewhere else. That separation is important when estimating cost and performance: the hosting plan runs the interface, but response speed and token charges can also depend on the external provider or model endpoint selected by the administrator.

Key Open WebUI features

Users can switch between connected models and keep conversations in one interface. Provider connections allow a team to avoid maintaining separate bookmarks and chat layouts for each endpoint, while administrators retain control over which services are exposed to the workspace.

The official feature set includes conversation tools, file attachments, web search options, and extensibility through APIs and tools. Some functions require additional services, credentials, or configuration, so the presence of a menu item does not mean the related provider is included with the hosting plan.

Open WebUI is not an AI model by itself. A working installation still needs an accessible provider or inference endpoint. That makes it suitable for teams consolidating existing AI access, but less suitable for someone expecting the hosting plan to include model compute or usage credits.

Who uses Open WebUI

Developers use Open WebUI to compare responses from different providers through a consistent chat experience. Small organisations can give staff a shared interface while controlling connection settings centrally, and technical teams can place a familiar web front end in front of an internal or external OpenAI-compatible service.

The product is most useful when the model strategy is already understood. Teams should decide which provider owns the inference workload, how API keys are governed, and whether uploaded material is appropriate for that provider before inviting a wider group.

Self-hosting Open WebUI: requirements and cost

Resource use on the interface is driven by active users, stored conversations, uploaded files, and optional processing performed inside the application. The catalogue marks the database requirement as “Not required”; the deployment keeps persistent application data without a separate PostgreSQL or MariaDB container.

AvaHost maps Open WebUI to Plan 2 at €5, while the software itself is free to run. The plan includes the hosted interface, automatic application updates, scheduled backups, custom domains, and automated HTTPS. It does not include model subscriptions, API usage, or inference hardware, and local server-side model use is blocked at the Docker level. Users must connect their own remote model provider or compatible endpoint.