Official training and fine-tuning package for the LTX video model, for training LoRAs for characters, effects or styles. Supports extension, inpainting/outpainting, audio and video-to-video workflows.

LTX Trainer
by Lightricks
Official Lightricks package for training and fine-tuning LTX-2 video models, including LoRAs for characters, effects and styles.
Updated EU status checked Last shown
LTX Trainer at a glance
- What is LTX Trainer?
- LTX Trainer is a library from Lightricks. Official Lightricks package for training and fine-tuning LTX-2 video models, including LoRAs for characters, effects and styles.
- Who makes LTX Trainer?
- LTX Trainer is made by Lightricks.
- Does LTX Trainer work in the EU?
- Available in the EU. The trainer is a self-hostable GitHub package with downloadable model weights, and no EU exclusion appears in the provided material. Weights are gated on Hugging Face, which requires accepting terms, and the license text was not reviewed. Last verified 14 Nov 2026.
- How much does LTX Trainer cost?
- Open source. Code is free to download from GitHub; requires own GPU compute. Model weights are gated on Hugging Face (accept terms).
- Where was LTX Trainer shown?
- In 1 creator video by AI Search. Most recently by AI Search on 21 Jun 2026, at 12:00 in “New robot waifus, GLM 5.2 craze, AI spas, new world models, new science agents: AI NEWS”.
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
LTX Trainer is the training and fine-tuning package in Lightricks' LTX-2 repository (packages/ltx-trainer). It is used to adapt the LTX audio-video diffusion model, for example by training LoRAs for specific characters, visual effects or styles.
According to creator coverage, it supports workflows such as video extension, inpainting/outpainting, audio, and video-to-video training. It sits alongside the repo's inference pipelines (distilled, DFR, retake and others) and the model weights hosted on Hugging Face.
It is run locally from the cloned repository (dependencies installed with uv), so it requires your own GPU hardware. The base model weights are large (the quick-start download is roughly 66 GiB) and are gated on Hugging Face, so the model terms must be accepted before download. The repository license was not shown in the provided material.
- LoRA training for characters, effects and styles
- Fine-tuning of LTX-2 audio-video diffusion models
- Supports extension, inpainting/outpainting and video-to-video workflows
- Audio-aware training
- Part of the open LTX-2 monorepo alongside inference pipelines
As seen on YouTube
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Similar AI tools
Closest neighbours by meaning, not just by category.
LTX 2.5
Model · Lightricks
Open-weights Lightricks video model with native audio, multi-shot generation, auto duration and up to 4K output.
Krea 2 LoRAs
App · Krea AI
LoRA training on Krea's platform to teach Krea 2 a custom style, character, face or product from a few images.
AI Toolkit
Library · Ostris
Open-source GUI and CLI training suite for fine-tuning image and video diffusion models, including LoRAs, on consumer GPUs.
MiniMax H3 Turbo LoRAs
Model · LightX2V
Few-step LoRA for MiniMax-H3 that cuts audio-video sampling from about 20 steps to 4–8 in ComfyUI.
Just Dub It
Model
Video dubbing method that adapts LTX 2.3 with a LoRA to translate speech and re-sync lips in a single model.
LTX Director
Library · WhatDreamsCost
ComfyUI custom node that works as a timeline video editor for LTX video generation, with keyframes, audio and IC-LoRA support.