Open UI - Open WebUI Client

Open UI is a beautiful, native iOS client for Open WebUI. Connect to your self-hosted server and chat with any AI model you've configured — right from your iPhone.

Built 100% in SwiftUI, Open UI delivers a fast, polished experience with features the browser can't match.

STREAMING CHAT WITH RICH MARKDOWN
Get real-time, word-by-word streaming responses. Messages render with full markdown — syntax-highlighted code blocks with copy buttons, tables, math equations, block quotes, headings, and more. Everything flows smoothly as it streams in with no layout jumps.

LIVE CROSS-DEVICE STREAMING
Start a generation on the website and watch it stream in real-time on the app, token by token. Regenerations, title updates, follow-ups, and completion events all sync instantly between devices.

CHART RENDERING
When the AI outputs chart data in code blocks (Chart.js, ECharts, Highcharts, Vega-Lite, or Plotly), it's automatically rendered as a beautiful native Swift Chart right in the conversation. Supports line, bar, area, scatter, bubble, pie, heatmap, and histogram charts. Includes a Chart/Source toggle to view the raw JSON data.

HTML PREVIEWS
HTML code blocks are now rendered as a live preview instead of raw code. Includes a Preview/Source toggle, fullscreen mode, copy button, share sheet, and dark mode support — all in a scrollable sandboxed webview.

VOICE CALLS WITH AI
Have a real voice conversation with AI. Open UI uses CallKit for native iOS call integration, with animated visualizations that react to voice, real-time transcription, and familiar call controls. It feels like a real phone call.

KNOWLEDGE BASES (RAG)
Type # in the chat to browse and attach knowledge collections, folders, and files from your server. Your AI responds using the context you provide — just like the web UI.

WEB SEARCH & CITATIONS
Toggle web search per message. Your AI searches the web and responds with numbered source citations, favicons, and tappable links.

TOOLS
Enable server-side tools per conversation. Tool calls render inline with expandable arguments, results, and status indicators.

REASONING / THINKING DISPLAY
See the full chain-of-thought from models like DeepSeek and QwQ. Collapsible "Thought for X seconds" blocks let you explore the reasoning process.

TEXT-TO-SPEECH
Choose from four TTS engines:
• Marvis Neural Voice — on-device AI voice powered by MLX
• Server