Locally Uncensored

Overview

Locally Uncensored is a free, open-source desktop AI studio that combines chat, coding, image generation, and video generation in one application. It installs like a normal desktop app without a command line or Docker, and its first-start wizard detects local AI engines and helps install suitable models. Local mode keeps chats on the user's computer, while LU Cloud is an optional hosted service for browser or phone use.

Key Features

• All-in-one desktop studio for chat, coding, image generation, and video generation in one interface.
• Installs like a normal app with no command line or Docker; the first-start wizard checks local AI engines and can download a suitable model.
• Chat supports local and optional cloud providers, file upload with vision, thinking mode, and 20+ provider presets.
• Create tab offers twelve image and video modes, including image, edit, cutout, enhance, erase, video, animate, character, lipsync, music, extend, and motion.
• Coding Agent reads project files, explains findings, and shows changes as a diff before applying them; it includes 14 MCP tools.
• Agent Mode can search the web, write files, and run steps for a goal, while Remote access lets a phone use the PC as the compute host.

Use Cases

• Private local chat and photo- or document-assisted Q&A without an account or usage limit after setup.
• Creating character art, images, and short video clips from text or an existing image.
• Diagnosing bugs and building small websites with the Coding Agent and Agent Mode.
• Using the desktop studio from a phone through QR code and passcode remote access.

Getting Started

• Download the Windows, macOS, or Linux installer and install it like a normal app; the app keeps itself up to date.
• On first start, let the wizard detect local AI engines and download a model recommended for the PC.
• Pick a model and begin in Chat, Code, or Create; an optional LU Cloud account can be added later in settings.

Deployment & Requirements

• Windows 10/11 x64; Linux x64 via AppImage, .deb, or .rpm.
• macOS is beta on Apple Silicon with macOS 14 or newer; Intel Macs are not supported.
• Text chat needs 8 GB RAM.
• Image generation needs an NVIDIA card, or an AMD card with ROCm on Linux, from 6 GB VRAM with the starter checkpoint and 8 to 10 GB for FLUX. Video generation starts at 6 GB for AnimateDiff and FramePack F1, 8 GB for the Wan 2.1 starter, and 12 GB or more for Wan 2.2 and

Before You Adopt

• License: AGPL-3.0. Review its terms before using, modifying, or distributing the project.
• The macOS desktop build is in beta and only supports Apple Silicon on macOS 14 or newer; Intel Macs are not supported.
• Voice is a beta feature and currently speaks English.
• Computer control is beta, off by default, and not fully tested on Mac.
• Remote access over the internet is relayed by Cloudflare and is not end-to-end encrypted, so Cloudflare could read traffic.