LocalMode UI
Local-first, offline AI UI primitives + blocks - run ML models (chat, RAG, vision, audio) entirely in the browser. No servers, no API keys.
- Free
- Price
- Sep 12, 2026
- Listed
Overview
LocalMode UI is a shadcn-style registry of local-first AI components for the browser. It offers more than 100 copy-owned React components and 36 installable blocks that run real models on-device, with no server and no API key, across chat, knowledge and RAG, vision, audio, photo, agents and privacy.
Components are installed with paths such as @localmode/ui/conversation/message. Blocks build on the LocalMode packages, which wrap runtimes including Transformers.js, WebLLM, wllama, MediaPipe, LiteRT and Chrome's built-in AI. The project describes itself as local-first by design and cloud-compatible by contract.
Features
- Conversation components: Messages, prompt input with attachments, reasoning, sources, inline citations, tool calls, tool approval, branches and agent step timelines.
- Local-first model UI: Model downloader, selector, loading panel, catalog cards, storage meter, device capability grid and browser compatibility card.
- Media and vision: Bounding box overlay, video canvas, before-and-after viewer, media dropzone and a
use-webcamhook. - Audio: Voice orb, voice picker, mic selector, streaming speech panel and synced transcript viewer.
- Results and privacy: Confidence badges, similarity meters, redacted text display, passphrase gate and vault item cards.
- DevTools: Inference queue monitor, event log viewer, pipeline run inspector and model cache table.
- On-device blocks: Chat, RAG chat, object detector, voice notes, image search, PII redactor, background remover and image captioner.
Tech stack
- React, Tailwind CSS, Radix UI, shadcn CLI registry
@localmode/*packages for Transformers.js, WebLLM, wllama, MediaPipe, LiteRT and Chrome AI
Registry
Install via the shadcn CLI registry:
npx shadcn@latest add @localmode/component-name