open-webui /open-webui
User-friendly AI Interface (Supports Ollama, OpenAI API, ...)
★ 158,745⑂ 23,082👁 684⑃ 12,968● 11,205▤ 181
+103
Daily
+567
Weekly
158,745
stars
23,082
forks
684
watchers
12,968
prs
11,205
issues
181
releases
README
Open WebUI Open WebUI is a home for AI: an extensible, easy-to-use, self-hosted platform built to run entirely offline. Connect local models through Ollama or use OpenAI-compatible APIs for local and cloud models. English | 简体中文 | 繁體中文 | 한국어 | Deutsch | Español | Français | Italiano | Dansk | 日本語 | Polski | Русский | Bosanski | العربية | Norsk | Português (Brasil) | ไทย | Türkçe | Українська | বাংলা | Ελληνικά | Tiếng Việt !GitHub stars !GitHub forks !GitHub watchers !GitHub repo size !GitHub language count !GitHub top language !GitHub last commit Explore Open WebUI → Installation Python bash With uv installed (macOS/Linux) DATA_DIR=~/.open-webui uvx --python 3.12 open-webui@latest serve Or install and run with pip (Python 3.12) pip install open-webui open-webui serve Then open http://localhost:8080. Docker bash 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 Then open http://localhost:3000. | Image tag | Includes | | --------- | ---------------------------------------------------------------------------------------- | | :main | Standard image with local AI dependencies | | :slim | Lightweight image using external services for embeddings, document extraction, and voice | | :cuda | Open WebUI with CUDA acceleration | | :ollama | Open WebUI bundled with Ollama | | :dev | Latest development build | Add these options to docker run before the image name as needed: | Option | Purpose | | ---------------------------------------------------- | ---------------------------------------------------------------------------------------- | | -e OLLAMA_BASE_URL=http://your-ollama-server:11434 | Connect to Ollama on another server | | -e OPENAI_API_KEY=your_secret_key | Set your OpenAI API key | | -e HF_HUB_OFFLINE=1 | Disable Hugging Face model downloads for offline use | | --gpus all | Enable Nvidia GPU access for :cuda or :ollama; requires the Nvidia Container Toolkit | See the Docker installation guide for setup instructions and configuration options. Desktop App Download the native desktop app for macOS, Windows, and Linux. Read the docs → Ecosystem - Open Terminal / Terminals (Enterprise): The action layer that turns Open WebUI int
Top issues
#21675enh: analytics103 comments · 2026-02-20#19313feat: User specific MCP headers46 comments · 2025-11-20#13464enh: notes46 comments · 2025-05-03#23339feat: Add model-level toggle to disable reinjecting reasoning/thinking into prompts (prevents <think> tag imitation and rendering/parsing failures)44 comments · 2026-04-02#8050enh: channels37 comments · 2024-12-25#12619feat: Different content extraction settings for chat and knowledge-base22 comments · 2025-04-08#23357feat: send audio and video directly to natively multimodal models instead of forcing STT transcription18 comments · 2026-04-02#5872feat: data sources15 comments · 2024-10-02
Discussions
#34480[Share] EveryApp remote MCP server: text to speech, redaction, PDF and media tools from a chat0 comments · 2026-11-15#34473feat: Mark as read in a chat's menu when the chat is unread0 comments · 2026-11-15#34472feat: open the model editor's Manage links in a tab beside the editor0 comments · 2026-11-15#34467Web Search and Image Generation icons not showing and not working when using + menu after upgrading to 0.13.3.1 comments · 2026-11-15#34466feat: a Reset under an Arena model's image, as the model editor has0 comments · 2026-11-15#34463feat: a Remove from folder option in a chat's Move menu0 comments · 2026-11-15#34445feat: show who owns or sent each Search result, as the sidebar does for shared chats0 comments · 2026-11-14#34444Add Hermes Agent to the terminal agent providers (one-line runner.py allow-list entry)1 comments · 2026-11-14
Latest release
v0.13.3 v0.13.3 2026-11-08Added - 🦾 Agents in automations. The automation editor now lets you choose a terminal and working folder, then select a coding agent available there. Scheduled runs use that saved terminal and folder, and check that the terminal and agent are still available before starting. Agent automations send their results to a chat; channel destinations remain available for ordinary models. Commit - 📂 Change terminal and folder during a chat. You can now choose a different terminal or working folder after sending messages, including in temporary chats. Both controls remain disabled while a reply is bein
相关报道 (6)
CSDNGitHub Trending速递:五大热门开源项目背后的技术趋势与选型实战2026-11-09CSDNGitHub热点精选方法论:本地模型、AI前端与开源知识库实践2026-11-09StorageReview.comOpen WebUI Desktop 实战:在 Windows 上运行本地 LLM2026-10-19麻省理工科技评论如果明天互联网消失:这个年轻人希望为全人类留下一套“文明重启”的AI火种2025-11-29OSCHINAOpen WebUI v0.6.0 发布,正式支持 MCP2025-04-0753AIDeepSeek R1,本地部署才是王道!支持WebUI2025-01-29
News & coverage (19)
Hacker NewsOpen WebUI: ChatGPT-Style WebUI for Ollama▲ 29 · 3 comments · 2024-02-18Hacker NewsOpen WebUI License Change▲ 4 · 2 comments · 2025-05-06Hacker NewsAnnouncing the New Era of 'open-webui'▲ 3 · 0 comments · 2024-02-18Hacker NewsOpen WebUI v0.9.0 adds desktop app with task scheduling▲ 2 · 0 comments · 2026-04-22Hacker NewsOpenWebUI is not open source anymore▲ 2 · 0 comments · 2025-05-08Hacker NewsOpen WebUI Website Temporarily Inaccessible▲ 2 · 2 comments · 2025-02-26shattered.ioOpen WebUI vs LibreChat vs AnythingLLM: 3 Picks [2026]2026-11-04noteOpen WebUI's :latest is not a stable version. 3 things to know before updating2026-11-03noteHow to use open-webui: An introductory guide to talking to local AI via your browser2026-09-29TechRadar on MSNHow to self-host your own AI agent2026-08-25