What You Can Do with Open WebUI
One interface for every AI model. Private, extensible, and built for teams.
Open WebUI replaces the patchwork of AI tools your team juggles daily - ChatGPT for writing, a separate app for image generation, another for document search, spreadsheets full of prompts nobody can find. Everything lives in one place: conversations, knowledge, tools, and the models that power them.
π¬ Chat & Conversationsβ
Talk to any model, switch mid-conversation, and keep your context intact.
Your conversations are the core of Open WebUI. Chat with Ollama, OpenAI, Anthropic, or any OpenAI-compatible provider from a single interface. Attach files, search the web, execute code, and let the AI use tools - all without leaving the chat.
| π Multi-model chats | Run two models side-by-side and compare responses |
| π File & image uploads | Attach documents, images, and code for the AI to analyze |
| π Web search | AI searches the web and cites sources in real time |
| π Code execution | Run Python directly in the browser or via Open Terminal |
| π Message queue | Keep typing while the AI responds - messages send automatically |
| π§ Memory | The AI remembers facts about you across conversations |
| ποΈ Folders, tags, pins | Organize conversations however you work |
| π€ Voice & audio | Speech-to-text, text-to-speech, hands-free voice & video calls, realtime voice calls |
| πΌοΈ Image generation | Create and edit images with GPT-Image, Gemini, ComfyUI, and more |
| β±οΈ Automations | Schedule prompts to run automatically on recurring schedules |
| β Task management | Models maintain structured task lists for multi-step workflows |
Explore chat features β Β· Audio β Β· Image Generation β
π Knowledge & RAGβ
Give your AI access to your documents - and let it find what matters.
Upload files, build knowledge bases, and let the AI retrieve exactly the information it needs. Choose between vector search (RAG) for large collections or full-content injection for precision. With native function calling, models autonomously search, browse, and synthesize across your entire knowledge base.
| π 13 vector databases | ChromaDB and PGVector (officially maintained), plus non-core integrations: Qdrant, Milvus, Elasticsearch, and more |
| π Hybrid search | BM25 + vector search with cross-encoder reranking |
| π 8 extraction engines | Tika, Docling, Azure, Mistral OCR, Datalab Marker, MinerU, PaddleOCR, custom loaders |
| π€ Agentic retrieval | Models search and read your documents autonomously |
| π Full context mode | Inject entire documents - no chunking, no guessing |
| ποΈ File browser with preview | Browse a knowledge base's files and folders and preview them in place |
π€ Models & Agentsβ
Wrap any model with custom instructions, tools, and knowledge to build specialized agents.
Create a "Python Tutor" that always uses your style guide. A "Meeting Summarizer" with your company's template. A "Code Reviewer" with your linting rules baked in. Every agent is a configuration wrapper - pick any base model, bind knowledge, tools, and a system prompt.
| 𧩠Model presets | System prompts, tools, knowledge, and parameters in one package |
| π·οΈ Dynamic variables | {{USER_NAME}}, {{CURRENT_DATE}} injected automatically |
| π§ Bound tools | Force-enable specific tools per model |
| π₯ Access control | Restrict models to specific users or groups |
| π Global defaults | Set baseline capabilities and parameters for all models |
| ποΈ Model controls | Let people pick from approved parameter presets, such as thinking effort, right in the chat |
| π Version history | Every saved configuration is kept, and any of them can be set live again |
π Notesβ
Write, think, and refine with AI by your side.
A dedicated workspace for content that lives outside of any single conversation. Draft with a rich editor, use AI to rewrite text in place, and attach notes to any chat for precise context injection - no chunking, no vector search.
| βοΈ Rich editor | Markdown and Rich Text with a floating formatting toolbar |
| π€ AI Enhance | Rewrite or improve selected text in place |
| π Context injection | Attach notes to any chat for full-fidelity context |
| π Agentic access | Models can search, read, and update notes autonomously |
π¬ Channelsβ
Where your team and AI think together, in real time.
Persistent, shared spaces where humans and AI models participate in the same conversation. Tag @gpt-5.6 to draft a plan, then tag @claude to critique it - your whole team sees both responses in one timeline.
| π€ @model tagging | Summon any AI model into the conversation on demand |
| 𧡠Threads & reactions | Replies, pins, and emoji reactions |
| π Access control | Public, private, group-based, and direct message channels |
| π§ AI channel awareness | Models search and synthesize across channels autonomously |
| 𧩠Apps | External services join channels, answer @mentions and post messages with buttons |
β‘ Open Terminalβ
Give your AI a real computer to work on.
Connect a real computing environment to Open WebUI. The AI writes code, executes it, reads the output, fixes errors, and iterates - all from the chat. Handles files, installs packages, runs servers, and returns results.
| π₯οΈ Code execution | Runs real commands and returns output |
| π File browser | Browse, upload, download, and edit files in the sidebar |
| π Website preview | Live preview of web projects inside Open WebUI |
| π Isolation optional | Docker container or bare metal |
Want the whole machine in a browser tab? See Open WebUI Computer, a separate Open WebUI project.
π Extensibilityβ
Add any capability with Python tools, MCP or OpenAPI servers.
Open WebUI is a platform, not a locked-down product. Write Python tools that run inside the chat. Connect external services via OpenAPI or MCP. Build functions that filter, transform, or route every message. Import community-built extensions with one click.
| π Tools & Functions | Python scripts that run in the chat - built-in code editor included, with version history and Tool Search for large tool sets |
| π MCP support | Native Streamable HTTP for Model Context Protocol servers |
| π OpenAPI servers | Auto-discover tools from any OpenAPI-compatible endpoint |
| π Skills | Markdown instruction sets that teach models how to approach tasks, with supporting files and version history |
| β‘ Prompts | Slash-command templates with typed input variables and versioning |
π Authentication & Accessβ
Control who gets in, what they can do, and how it connects to your identity stack.
Open WebUI is multi-user from day one. Define roles, create user groups, set per-model access, and integrate with your identity provider. From a solo install to an organization with thousands of seats.
| π₯ RBAC | Roles, groups, and per-resource permissions |
| π SSO/OIDC/LDAP | Federated authentication with any identity provider |
| π SCIM 2.0 | Automated user and group provisioning |
| π API Keys | Programmatic access for scripts, bots, and integrations |
| π± Two-Factor Sign-In (MFA) | Require a code from an authenticator app at every sign-in |
Learn about Authentication & Access β
π§ Administrationβ
Monitor usage, evaluate models, and communicate with your users.
| π Analytics | Usage dashboards with message volume, token consumption, and cost tracking |
| π Evaluation | Model arena, A/B testing, and ELO-based leaderboards |
| π’ Banners | Customizable system-wide announcements |
| ποΈ Default Interface Settings | Instance-wide starting values for everyone's interface options |
| π Webhooks | Notifications for sign-ups, chat completion, and external integrations |
Learn about Administration β
ποΈ Deploy Anywhereβ
Stateless architecture for horizontal scaling, on Docker, Kubernetes, pip, or bare metal. With data in shared databases and storage, and Redis coordinating instances, add application replicas behind a load balancer as demand grows.
| π³ Docker & Compose | One-command deploy with GPU support |
| βΈοΈ Kubernetes & Helm | Production-ready orchestration |
| π¦ pip install | pip install open-webui && open-webui serve |
| βοΈ Cloud storage | S3, GCS, Azure Blob for stateless instances |
| π OpenTelemetry | Traces, metrics, and logs to your observability stack |
| βοΈ Horizontal scaling | Stateless application replicas with shared databases, storage, and Redis |
