Discover · GitHub · LTX 2

Video production with artificial intelligence in the local system

Developed by Lightricks, LTX-2 offers a Python inference and low-rank adaptation (LoRA) training package for artificial intelligence models that produce audio and video. This toolset allows users to train LTX-2 models with their own data and run model outputs on local systems.

★ 9,267
GitHub Trending · 2026-06-19

Updates

August 27, 2026: Stars 8,587 → 9,267, latest release v1.3.0 (August 26, 2026).
August 12, 2026: Stars 8,554 → 8,587, latest release v1.2.0 (August 11, 2026).
August 10, 2026: Stars 7,550 → 8,554.

What you get

Provides audio and video synchronization
You can train LoRA with your own data
High quality video production on local system

Installation

Clone the repository from GitHub and enter the directory

git clone https://github.com/Lightricks/LTX-2.git
cd LTX-2

Download model weights (Hugging Face CLI)

hf download Lightricks/LTX-2.3 ltx-2.3-22b-distilled-1.1.safetensors --local-dir models/ltx-2.3

Running it

run inference pipeline with uv

uv run python -m ltx_pipelines.distilled --distilled-checkpoint-path models/ltx-2.3/ltx-2.3-22b-distilled-1.1.safetensors

If you don't write code

🤖 Paste this into your AI agent (Claude Code · Codex · Antigravity)

Please create a video using the LTX-2 model that describes the scene I want in detail and includes audio and video synchronization. Have the model produce output by specifying scene details, character's appearance, camera angle and speech text.

Related dictionary terms

Who it is forFor users who want to create audio and video AI videos or train models on their own local system.

Links

TreScout did not build this tool · we found it in GitHub trends and wrote it up. This page describes the repository as of 2026-06-19: The star count and our text belong to that day, the repository may have changed since. Check the repository link for the current state. This page was machine-translated from the Turkish original · the Turkish version prevails.