Open WebUI: The ChatGPT alternative for your team – on your own servers

Open WebUI is a ChatGPT-style user interface that you on your own serversoperate. Your team chats with an AI as usual — only the conversations remain within your infrastructure. Open WebUI connects to local models (for example via Ollama) or with cloud services and offers user management, permissions and chat with your own documents. An important point upfront: Since 2025, the license requires from 50 users visible Open WebUI branding or a paid enterprise license.

What Open WebUI can do

• Familiar chat interface: conversation history, folders, templates – no need for employees to adjust.
• Multiple models: local open models and, if desired, cloud models in one interface.
• Chat with documents: Upload files or create knowledge collections on which the AI bases its answers – via RAG.
• Users and groups: Roles, permissions and sharing settings that determine which group may use which model and which knowledge.
• Login via single sign-on: Connection to central user management systems such as Authentik or Keycloak.
• Expandable: Tools and functions can be added, for example integration with automations.

Why companies want their own ChatGPT alternative

In many companies, employees are already using private AI accounts – often with company data and without rules (“shadow AI”). A dedicated interface creates an official, secure way: the company determines which models are available, where data is processed, and who has access. At the same time, employees keep the interface they are familiar with.

The licensing catch: branding from 50 users

Open WebUI was originally released under the permissive BSD license. Since version 0.6.6 (April 2025), a separate Open WebUI License: the name and logo of Open WebUI may not be removed or altered — unless the installation has at most 50 users in 30 days or you purchase a Enterprise License. Due to this clause, the license is no longer considered open source within the meaning of the Open Source Initiative. For most SMEs, this is not critical – but anyone who wants to operate the interface in their own design and with many users should plan for this.

Privacy in focus

• Local models: Inputs and documents remain on your server.
• Cloud models:If cloud services are connected as well, the inputs flow there too – with a DPA and clear rules on which data is permitted.
• Conversation histories:Define how long they are stored and who may view them.
• Logging and permissions:Who is allowed to use which knowledge collection?

Introduction in five steps

1. Goals and rules: What should the internal chat be used for? What is prohibited? Help: Create AI policy.
2. Choose models: one or two local models, if needed a cloud model for non-critical tasks – see Open-source AI models at a glance.
3. Operation: servers in Germany with sufficient graphics memory, login via Single Sign-On.
4. Knowledge collections: provide initial documents for a department.
5. Training:a brief introduction to how to ask good questions and check answers – the EU AI Act requires sufficient AI literacy.

Alternatives

In addition to Open WebUI, there are other self-hostable interfaces. Anyone already using Nextcloud can also use the integrated assistant there. For working with large document collections, a specialized AI document search.

bettersorted relies on open, transparent components – operated on servers in Germany and bundled in the automaisa Hub. We provide vendor-neutral advice, are a BAFA-registered consultant and authorized INQA coach. The consulting and guided implementation can be funded through the INQA-Coaching with 80% funding; the suitable path is shown by the Funding Check. For a non-binding initial consultation: Contact.

Example: municipal administration with an internal AI chat

An illustrative scenario: A municipal administration wants to use AI for draft texts, summaries of meeting documents, and searching bylaws. Open WebUI runs on a server in Germany, and login is handled via the existing directory through single sign-on. There are two knowledge collections – “Bylaws” for everyone, “HR” only for the HR department. Cloud models are disabled. After the training, a service directive defines which tasks are permitted.

Rules you should define

• Which data may be entered – and which must never be entered?
• Which models are available to which groups?
• How long are conversation histories stored?
• Who maintains the knowledge collections?
• How are AI outputs checked before they are shared externally?

A template for such rules is provided by Create AI policy.

Costs and funding

The costs of an in-house AI chat consist of three parts: Setup (planning, installation, integration, testing), ongoing operations (server or hosting, updates, monitoring, data backup) and support for employees (training, rules, contacts). We do not quote fixed prices because scope and starting conditions vary greatly. The software itself does not incur license costs for open-source solutions.

Funding is not for the technology, but for the consulting and guided implementation: Through INQA-Coaching the federal government covers 80% of coaching costs nationwide (up to €11,520, vouchers until 30/06/2028). A preliminary analysis can be funded through the BAFA consulting subsidy subsidize – with 80% in the new federal states, Lüneburg and Trier, otherwise 50%, for applications submitted by 31.12.2026. For investments, some states offer their own programs – see Funding opportunities for companies in Germany.

Frequently asked questions

Is Open WebUI free of charge?

For most companies, yes. However, the license requires that the Open WebUI branding remain visible once more than 50 users use the installation within 30 days – or a paid enterprise license.

Is Open WebUI open source?

The code is public, but the license in force since 2025 includes a branding clause. It therefore does not meet the criteria of the Open Source Initiative.

Can Open WebUI be connected to ChatGPT?

Yes. In addition to local models, cloud models can also be connected via their interfaces. In that case, the privacy terms of the respective provider apply to those requests.

Do I need a graphics card for Open WebUI?

Open WebUI itself does not. However, the language model it works with needs a graphics card with sufficient memory for good speed — or a model service in Germany.

As of October 2026. Models, versions and licenses change quickly – check the current license with the provider before making a decision.