# LTX-2

Official Python inference and LoRA trainer package for the LTX-2 audio–video generative model.

Repository: https://github.com/Lightricks/LTX-2
Canonical: https://ross.abutalabs.com/products/ltx-2
Homepage: https://ltx.io
Language: Python
License: NOASSERTION
License Family: other
Topics: generative-ai, ltx, ltx-2
Last push: 2026-08-26T10:47:10+00:00
Link (homepage): https://ltx.io
Link (site_page): https://docs.ltx.io/pricing
Link (site_page): https://ltx.io/studio/pricing
Link (site_page): https://ltx.io/about-us

## Health v2 (maintenance only)
Score: 83/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 99, longevity 17
- inputs: {"age_days": 242, "days_push": 7, "days_rel": 7, "gap_med": 14, "n_releases_24m": 2}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 9260, forks 1458 (observed 2026-08-28T04:10:29.108748+00:00)

## What it is
Official Python package from Lightricks providing inference pipelines and LoRA training for LTX-2/LTX-2.5, an open-weights DiT-based foundation model that generates synchronized audio and video (up to 4K HDR) from text, image, or audio prompts. It includes a companion ComfyUI integration and falls back gracefully across CUDA, Triton, and eager backends.

## Use cases
- generate video with synchronized audio from a text prompt
- run image-to-video generation locally on my own GPU
- fine-tune the LTX-2 video model with LoRA on my own footage
- generate 4K HDR video with an open-weights model
- use LTX-2 in ComfyUI workflows
- extend or edit existing video clips with a generative model
- self-host video generation to keep IP in-house

## When to choose
- you want open-weights, self-hostable audio-video generation with full control over fine-tuning and deployment
- you need production-grade video output including native 4K HDR and multishot consistency
- you want a ComfyUI-integrated video model or LoRA trainer in Python

## When to avoid
- you only need a hosted API without managing GPU infrastructure (though an API exists, the repo targets local inference)
- you lack a capable GPU - high-fidelity video diffusion models are compute-heavy
- you need simple image-only generation rather than audio-video

## Facets
- artifact type: library
- maturity: active
- function: llm-inference, machine-learning, deep-learning, video-processing, audio-processing, llm-training
- domain: artificial-intelligence, deep-learning, media, gpu-computing
- platform: python, windows, cross-platform
- tags: video-generation, audio-video-generation, diffusion-transformer, lora-training, open-weights, generative-ai, text-to-video, image-to-video, comfyui, video, audio, gpu, linux, macos

## Member repositories
- Lightricks/LTX-2 (main) score 83
- Lightricks/ComfyUI-LTXVideo (plugin) score 66

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:29.108748+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:22:38.936253+00:00, confidence not recorded.
  - readme: https://github.com/Lightricks/LTX-2 (fetched 2026-08-28T04:10:29.108748+00:00, sha 051e236d9057)
  - homepage: https://ltx.io (fetched 2026-08-29T08:22:53.972714+00:00, sha 70a70954219b)
  - site_page: https://docs.ltx.io/pricing (fetched 2026-08-29T08:22:53.981902+00:00, sha 692a428bba47)
  - site_page: https://ltx.io/about-us (fetched 2026-08-29T08:22:53.986038+00:00, sha f2d5eed47c6f)
  - site_page: https://ltx.io/studio/pricing (fetched 2026-08-29T08:22:53.983694+00:00, sha 5fcdbed13272)
- Data as of 2026-08-30T08:39:29.467469+00:00.