Instructions to use Lightricks/LTX-2.5 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Lightricks/LTX-2.5 with Diffusion Single File:

# No code snippets available yet for this library.

# To use this model, check the repository files and the library's documentation.

# Want to help? PRs adding snippets are welcome at:
# https://github.com/huggingface/huggingface.js

LTX-2

How to use Lightricks/LTX-2.5 with LTX-2:

# Install the LTX-2 pipelines
git clone https://github.com/Lightricks/LTX-2.git
cd LTX-2
uv sync --extra natten

# Download weights from this repo
# Substitute filenames from this repo's "Files and versions" if they differ
hf download Lightricks/LTX-2.5 \
    diffusion_models/<distilled-transformer>.safetensors \
    text_encoders/gemma4-12b-with-proj-ltx-2.5-bf16.safetensors \
    vae/<video-vae>.safetensors \
    vae/<audio-vae>.safetensors \
    latent_upscale_models/<spatial-upsampler>.safetensors \
    latent_upscale_models/<temporal-upsampler>.safetensors \
    --local-dir models/LTX-2.5
# DFR requires the detailing IC-LoRA (separate repo; strength is fixed at 0.5)
hf download Lightricks/LTX-2.5-22b-IC-LoRA-Pixel-Spatial-Upscaler --local-dir models/LTX-2.5-22b-IC-LoRA-Pixel-Spatial-Upscaler

# Distilled LTX-2.5 pipeline (fast)
uv run python -m ltx_pipelines.distilled \
    --transformer-path models/LTX-2.5/diffusion_models/<distilled-transformer>.safetensors \
    --text-encoder-path models/LTX-2.5/text_encoders/gemma4-12b-with-proj-ltx-2.5-bf16.safetensors \
    --video-vae-path models/LTX-2.5/vae/<video-vae>.safetensors \
    --audio-vae-path models/LTX-2.5/vae/<audio-vae>.safetensors \
    --spatial-upsampler-path models/LTX-2.5/latent_upscale_models/<spatial-upsampler>.safetensors \
    --num-frames 121 \
    --prompt "A beautiful sunset over the ocean" \
    --output-path output.mp4
# For image-to-video, add: --image path/to/image.jpg 0 0.8

# DFR pipeline (higher detail fidelity; optional temporal 2x/4x)
uv run python -m ltx_pipelines.dfr_pipeline \
    --transformer-path models/LTX-2.5/diffusion_models/<distilled-transformer>.safetensors \
    --text-encoder-path models/LTX-2.5/text_encoders/gemma4-12b-with-proj-ltx-2.5-bf16.safetensors \
    --video-vae-path models/LTX-2.5/vae/<video-vae>.safetensors \
    --audio-vae-path models/LTX-2.5/vae/<audio-vae>.safetensors \
    --spatial-upsampler-path models/LTX-2.5/latent_upscale_models/<spatial-upsampler>.safetensors \
    --temporal-upsampler-path models/LTX-2.5/latent_upscale_models/<temporal-upsampler>.safetensors \
    --detailing-lora models/LTX-2.5-22b-IC-LoRA-Pixel-Spatial-Upscaler/ltx-2.5-22b-ic-lora-pixel-spatial-upscaler-x2-1.0.safetensors \
    --spatial-upscalings 1 \
    --temporal-upscalings 1 \
    --height 1088 \
    --width 1920 \
    --num-frames 121 \
    --prompt "A beautiful sunset over the ocean" \
    --output-path output.mp4
# For 4K: --spatial-upscalings 2 --width 3840 --height 2176
# For image-to-video, add: --image path/to/image.jpg 0 0.8

Notebooks
Google Colab
Kaggle

LTX 2.5 Just Changed Local AI Video Forever!

#37

by LabMike3D - opened Aug 14

Discussion

LabMike3D

Aug 14

Hey everyone!

Here is a quick demo for those who haven't had a chance to test LTX 2.5 yet. LTX 2.5 can do incredible things, but keep in mind that it’s not always smooth sailing. Get ready to invest a ton of time before you start digging out high-quality and, most importantly, consistent results. Based on my long experience using LTX, I know very well that it can often be a hit-or-miss experience.

In the beginning, you will waste a lot of generations, and you might feel like complaining (just like a lot of people do) that LTX is trash and MiniMax is simply the best. Which is, of course, nonsense. Every model has its strengths and weaknesses. I could write a whole essay about it here, but I won't. Instead, I’ll gradually add more samples so everyone can see (and hear) for themselves what the new LTX 2.5 can really do.

For me personally, a huge deal right now is that LTX 2.5 can generate quick iterations even on a fairly mid-range GPU like the RTX 3060 12GB. And that is absolutely crucial for me. Tuning a single scene often takes 10 to 15 iterations and hours of optimization, making it a pretty lengthy process. But hey, every hobby has its thing. Some people go fishing and stare at the water for hours waiting for a bite, while others spend their evenings playing with ComfyUI and learning new skills.

If you are in the same boat, let me know in the comments what you are currently working on in ComfyUI! Oh, and last but not least, a quick technical note. Although I’ve managed to reuse my older workflows and prompts from LTX 2.3 in LTX 2.5, the results are practically unusable so far. This is especially true for workflows where I use external audio for avatars. However, it will get sorted out and optimized over time. I'm sure Kijai and RuneXX will put their heads together and build some F-tuned WF's for LTX 2.5 soon.

Cheers!

LTX 2.5 Just Changed Local AI Video Forever!

https://youtu.be/OHTw7l_5aYE