Open Source Release by GeethuDinoYT v1.0.5
Latest Update: Improved stability and performance

AI Video Generation
On Any GPU

The unrestricted, open-source AI video editor that runs on 4GB+ VRAM. No API keys. No forced cloud services. Fully offline. Generate unlimited videos locally on your existing hardware.

View on GitHub 🐙
4GB+
VRAM Required
100%
Offline
Zero
API Keys
Generations

Why Unrestricted?

Unlike the official release that forces API usage on lower-end hardware, this optimized version runs fully local on virtually any modern GPU.

Official LTX Desktop

Requires high-end hardware

  • Requires 16GB+ VRAM for local generation [^5^]
  • Forces API mode if VRAM < 16GB [^5^]
  • macOS only supports API mode (paid) [^5^]
  • Requires API key for cloud generation
  • Potential usage costs

Unrestricted Version

Optimized by GeethuDinoYT

  • Works on 4GB+ VRAM GPUs
  • Never forces API - always local
  • macOS & Windows & Linux native support
  • Zero API keys required
  • Fully free, unlimited generations

Note: This is a community-optimized build based on the open-source LTX-Desktop repository [^5^]

Suggest an update

What should we build next?

Share your improvement ideas, feature requests, or performance thoughts. Your feedback helps shape the next unrestricted release.

Submit Your Suggestion

See It In Action

Powered by LTX-2.3 with synchronized audio generation and 4K support.

Desktop Interface

Non-linear editing with AI generation built-in

Local Generation

Runs entirely on your GPU - no cloud needed

4GB VRAM Support

Optimized quantization and memory management allow running on GTX 1650, RTX 3050, and other entry-level GPUs.

Fully Offline

No internet connection required after initial setup. Generate unlimited videos without API keys or cloud services.

Unlimited Generations

No usage caps, no watermarks, no subscription fees. Generate hundreds of clips overnight locally.

Professional Features

Everything from the official LTX Desktop, optimized to run on your hardware.

Text-to-Video

Generate videos from text prompts with precise control over style and composition [^1^].

Image-to-Video

Animate static images with motion and depth. Multi-keyframe conditioning support [^5^].

Video-to-Video

Transform existing videos with prompt guidance while maintaining motion structure [^1^].

Audio Sync

LTX-2.3 generates synchronized audio and video in a single pass [^1^].

Timeline Editor

Non-linear editing with ripple cuts, trimming, adjustment layers, and XML export [^1^].

LoRA Support

Load custom LoRA weights for specialized styles and consistent characters [^1^].

4K Native

Generate up to 4K resolution at 50fps with spatial upscaling support [^1^].

Regenerate Shot

Right-click any clip to regenerate with variations while keeping previous versions [^1^].

System Requirements

Optimized to run on virtually any modern GPU. No high-end hardware required.

GPU

4GB+

NVIDIA GTX 1650 or better
AMD support coming

Storage

160GB

SSD recommended
For model weights

RAM

16GB

32GB recommended
For 4K generation

Compatibility Note: Unlike the official version that requires 16GB VRAM and forces API usage on lower hardware [^5^], this unrestricted build uses optimized quantization to run on 4GB+ GPUs fully offline. Windows 10/11 and Linux supported. macOS (Apple Silicon) version available with local generation support.

Download Unrestricted

Choose your platform. Version 1.0.5 includes stability improvements and enhanced performance. All versions include full local generation support with no API requirements.

Recommend a feature for the next release

Your feedback helps make the unrestricted version stronger.

Send Suggestion

Windows

4GB+ VRAM

Windows 10/11 (64-bit)
NVIDIA GPU with CUDA

v1.0.5 Unrestricted • 185 MB

Linux

4GB+ VRAM

Ubuntu 22.04+
NVIDIA GPU with CUDA

v1.0.5 Unrestricted • 210 MB

macOS

Local Gen

macOS 13+ (Ventura)
Apple Silicon M1/M2/M3

v1.0.5 Unrestricted • 195 MB
Open Source
Apache 2.0 License
GitHub Available

How It Works

Three simple steps to unlimited AI video generation.

1

Download & Install

Get the unrestricted build for your platform. Install like any normal application.

2

Download Models

First launch downloads ~150GB of AI models automatically. No manual configuration needed.

3

Generate Offline

Start generating immediately. No API keys, no internet required, no usage limits.