Open WebUI
A self-hosted chat and knowledge interface for local and cloud model providers.
By Open WebUI · reviewed 2026-08-07
Deep profile checked 2026-08-16 · Open WebUI documentation and versioned container releases
What Open WebUI does
Open WebUI adds conversations, users, knowledge collections, tools, and administration around inference providers such as Ollama and OpenAI-compatible APIs. It is primarily an application layer, so model performance and hardware support depend on the connected runtime.
Pricing model
Source available; license conditions apply
Position in the stack
Where Open WebUI fits
Open WebUI is the user-facing application layer. It connects to Ollama and OpenAI-compatible providers, then adds conversations, users, knowledge collections, tools, and administration. The connected runtime still determines which models fit the hardware and how quickly they run.
Typical deployment flow
A practical sequence for evaluating Open WebUI before making it part of a permanent stack.
- STEP 1
Provide inference
Run Ollama or another compatible API on the same machine, another server, or an approved cloud provider.
- STEP 2
Deploy the interface
Install with Docker, Kubernetes, Python, or the supported desktop packaging and configure persistent storage.
- STEP 3
Add users and knowledge
Configure provider access, permissions, document collections, tools, and any network exposure controls.
Best for
- Adding a shared UI to Ollama or an API server
- Private document chat
- Multi-user local AI access
Not the right layer for
- Replacing an inference engine
- Scheduling GPUs or optimizing model kernels
System fit
Requirements and hardware notes
Application requirements are separate from the memory needed by the selected model and context window.
Check your hardwareMinimum baseline
- macOS, Linux, or Windows
- Docker, Kubernetes, or supported Python
- A reachable Ollama, OpenAI-compatible, or Open Responses provider
Recommended baseline
- Docker with a named persistent volume
- A pinned stable image tag
- Python 3.11 for non-container production installs
Hardware-specific notes
- Open WebUI itself is primarily an application layer; inference hardware belongs to the connected runtime.
- The CUDA image is required when bundled local features need NVIDIA acceleration.
- The slim image downloads some supporting models on first use.
Installation and deployment paths
Choose one path that matches the number of users and the level of operations you can maintain.
Data boundary
What stays local and what may leave
- Local data
- Chats, users, configuration, and uploaded knowledge live in the configured data volume.
- Network dependencies
- Prompts go to each configured model provider. Local Ollama stays local; cloud providers do not.
- Accounts and telemetry
- Account mode, web search, tools, and external providers each add separate trust boundaries.
License checkpoints
- Review the current Open WebUI license and branding terms before commercial redistribution.
- Connected model and provider terms remain independent.
Operational checkpoints
- Persist WEBUI_SECRET_KEY and the data volume before upgrades.
- Single-user mode cannot be switched back to multi-account mode after initial configuration.
- Do not share data volumes between development and stable images.
Capabilities
Capabilities refer to the tool's application layer. Hardware and model support can still depend on a connected inference engine.
Deployment and compatibility
- Deployment
- Desktop / localSelf-hosted
- Platforms
- macOS · Windows · Linux · Docker · Kubernetes
- Hardware backends
- Uses connected model runtime
- Install methods
- Docker · Kubernetes · Python package · Desktop application
Models and integrations
- Model formats
- Uses connected model runtime
- Common integrations
- OllamaOpenAI-compatible APIsvLLMMCP servers
Strengths
- Broad provider compatibility
- Strong chat, knowledge, and user-management features
- Easy Docker deployment
Limitations
- Requires a separate inference provider for most models
- Branding and redistribution conditions differ from an OSI license
Planning checklist
Before you choose Open WebUI
Answer these questions with the exact models, hardware, users, and data you expect to operate.
- Which inference endpoints will users be allowed to access?
- Where will conversations, uploads, and embeddings be stored?
- Do license and branding conditions fit the intended distribution?
- How will authentication, backups, and upgrades be managed?
Open WebUI FAQ
Does Open WebUI run models itself?
Its main role is the application interface. Inference normally comes from Ollama, an OpenAI-compatible server, or another configured provider.
Can Open WebUI be used by a team?
Yes. It includes multi-user and administrative features, but secure external access still requires careful authentication, networking, storage, and backup configuration.
Is Open WebUI open source?
Its source is available, but the current Open WebUI License includes conditions that differ from a standard OSI-approved open-source license.
Official sources
Use these links to confirm current compatibility and installation requirements.
Related tools
A desktop and self-hosted workspace for document chat, RAG, and AI agents.
A self-hostable personal AI and knowledge assistant with document search, agents, automations, and research workflows.
A self-hosted multi-provider chat application with agents, files, search, MCP, and model routing.
Comparisons featuring Open WebUI
Use a reviewed comparison when the choice is between two adjacent tools.
Open WebUI is a flexible multi-model chat interface, while AnythingLLM is organized around document-backed workspaces and ready-made knowledge workflows.
LibreChat and Open WebUI are self-hosted chat and agent interfaces that connect to local or hosted providers, but their deployment dependencies, licensing, and feature emphasis differ.
Execution evidence
Known working recipes using Open WebUI
Recipes connect hardware, a model artifact, tools, settings, verification, and a reportable result.