Open WebUI is a self-hosted AI platform positioned as the "home of AI," capable of running fully offline, supporting Ollama and OpenAI-compatible APIs, and providing users with a unified interface across local and cloud models.
Main Capabilities
Model and API Integration
Supports connecting to any OpenAI-compatible API and local Ollama models, and can point API URLs to LMStudio, GroqCloud, Mistral, OpenRouter, vLLM, etc., freely mixing and matching different providers.
Permissions and Management
Provides fine-grained RBAC and user groups. Administrators can define roles, groups, and permissions, secure by default, and can customize the experience for different groups.
Plugins and Extensions
Supports plugin extensions such as Filters, Actions, Pipes, Tools, and Skills, and can connect to external services through MCP, MCPO, and OpenAPI tool servers to build custom integrations, rate limiting, approval flows, data connections, and more.
Models and Agents
Any base model can be wrapped into a specialized agent with custom instructions, tools, and knowledge, supporting dynamic variables, per-user/group access control, and the ability to import community presets through Open WebUI Community.
Terminal and File System Execution
Open Terminal provides agents with a terminal and file system to execute multi-step tasks, analyze data, run scripts, fix errors, and produce files directly in the conversation. Terminals (Enterprise) supports per-user isolated environments, resource limits, and automatic lifecycle management.
Notes, Channels, and Memory
Notes provides an independent workspace where you can draft with a rich editor, rewrite selected text with AI, and attach notes to any conversation. Channels are real-time shared spaces for teams and AI models to collaborate, supporting threads, reactions, pinning, and access control. Persistent memory lets AI remember user facts across conversations and continue context.
Workflows and Automation
Displays in real time the process of AI building and completing checklists, and can queue messages and send them automatically while AI is responding. Built-in personal and shared calendars support month/week/day views, recurring events, color coding, participants, and reminders, and models can manage schedules conversationally through native function calling. Automations supports running prompts on recurring schedules, with run records shown in the calendar and linked back to the generated conversation.
Interaction andMultimedia
Responsive design with PWA support for desktop, laptop, and mobile, with offline access on localhost. Full Markdown and LaTeX support. Integrated voice and video calls, supporting multiple speech-to-text providers (local Whisper, OpenAI, Deepgram, Azure) and text-to-speech engines (Azure, ElevenLabs, OpenAI, Transformers, WebAPI).
Storage and Data
Built-in key-value storage API for persisting artifacts, supporting tools such as logs, trackers, and leaderboards for personal and shared data scopes.
Local RAG and Search
Retrieval-augmented generation is supported by 9 vector databases and multiple content extraction engines (Tika, Docling, Document Intelligence, Mistral OCR, PaddleOCR-vl, external loaders), supporting hybrid search (BM25 + vector) with reranking and full-context mode. Documents can be loaded into conversations or pulled from the library via the # command. Web search supports dozens of providers, including SearXNG, Google PSE, Brave Search, Kagi, Mojeek, Tavily, Perplexity, Firecrawl, serpstack, serper, Serply, DuckDuckGo, SearchApi, SerpApi, Bing, Jina, Exa, Sougou, Azure AI Search, and Ollama Cloud, and injects results directly into conversations.
Image Generation and Editing
Supports creating and editing images with multiple engines such as OpenAI DALL·E, Gemini, ComfyUI (local), and AUTOMATIC1111 (local), with support for generation and prompt-based editing.
Multi-Model and Analytics
Chat with multiple models simultaneously, leveraging their respective strengths in parallel. The admin dashboard tracks message volume, token consumption, and cost for users and models, with built-in arena, A/B testing, and ELO leaderboards to evaluate models.
Database and Storage
Choose SQLite (with optional encryption) or PostgreSQL, and files can be stored locally, on S3, Google Cloud Storage, or Azure Blob Storage. Supports 9 vector databases: ChromaDB, PGVector, Qdrant, Milvus, Elasticsearch, OpenSearch, Pinecone, S3Vector, and Oracle 23ai.
Enterprise Features
Full LDAP/Active Directory integration, SSO via trusted headers and OAuth providers, and SCIM 2.0 automated provisioning (Okta, Azure AD, Google Workspace, etc.). Native support for Google Drive and OneDrive/SharePoint file selection. Built-in OpenTelemetry support for tracing, metrics, and logs. Redis-based session management and WebSocket support for multi-worker, multi-node load-balanced deployments. Multilingual i18n support.
Installation Methods
Python pip Installation
Python 3.11 is required to avoid compatibility issues:
pip install open-webui
Start:
open-webui serve
Visit http://localhost:8080
Docker Quick Start
When installing with Docker, be sure to include -v open-webui:/app/backend/data to correctly mount the database and prevent data loss. If you need to include Ollama or CUDA acceleration, it is recommended to use the official image with the :cuda or :ollama tag. Enabling CUDA requires installing the Nvidia CUDA container toolkit on a Linux/WSL system.
Default configuration (Ollama on this machine):
docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:main
Ollama on another server:
docker run -d -p 3000:8080 -e OLLAMA_BASE_URL=https://example.com -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:main
Nvidia GPU support:
docker run -d -p 3000:8080 --gpus all --add-host=host.docker.internal:host-gateway -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:cuda
Using only the OpenAI API:
docker run -d -p 3000:8080 -e OPENAI_API_KEY=your_secret_key -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:main
Bundled Ollama support (GPU):
docker run -d -p 3000:8080 --gpus=all -v ollama:/root/.ollama -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:ollama
Bundled Ollama support (CPU only):
docker run -d -p 3000:8080 -v ollama:/root/.ollama -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:ollama
After installation, visit http://localhost:3000。其他安装方式包括非 Docker native installation, Docker Compose, Kustomize, and Helm.
Troubleshooting
If connection issues occur, it is usually because the WebUI Docker container cannot access the Ollama server at 127.0.0.1:11434 (host.docker.internal:11434) inside the container. You can use --network=host The flag is resolved; the port will change from 3000 to 8080, i.e. http://localhost:8080。
Update and Development Branches
The update guide is in the official documentation.:dev The branch contains the latest unstable features and changes, and may have bugs or incomplete features; use at your own risk. When running Open WebUI in an offline environment, you can set the HF_HUB_OFFLINE environment variable to 1 to prevent downloading models from the internet.
Ecosystem
Open WebUI is the core, surrounded by companion applications and infrastructure: Open WebUI Computer is an independent, mobile-first computer and coding agent that runs on your own machine, with files, terminal, and git accessible in a browser tab, and can also be connected into Open WebUI as a model, or accessed from Telegram, WhatsApp, etc. Open Terminal and Terminals (Enterprise) are self-hosted computing environments that provide AI with a place to write code, run it, read output, fix errors, and iterate. Terminals provides per-user isolated containers, independent credentials, resource limits, and network rules, with automated lifecycle management on Docker or Kubernetes. oikb can sync knowledge bases from 45+ sources (GitHub, Confluence, ServiceNow, Salesforce, Jira, Slack, SharePoint, Notion, etc.). The native desktop app supports macOS, Windows, and Linux, offering a system-level Spotlight chat bar, screenshot capture, push-to-talk voice, and fully local inference via the built-in llama.cpp engine.
License
This project contains code under multiple licenses. The current codebase includes components under the Open WebUI License (with the additional requirement to preserve the "Open WebUI" branding), as well as contributions from previous contributors under their respective licenses.
