<h1 id="ltx-2">LTX-2</h1>This model card focuses on the LTX-2 model, as presented in the paper <a target="_blank" rel="ugc" href="https://huggingface.co/papers/2601.03233">LTX-2: Efficient Joint Audio-Visual Foundation Model</a>. The codebase is available <a target="_blank" rel="ugc" href="https://github.com/Lightricks/LTX-2">here</a>.LTX-2 is a DiT-based audio-video foundation model designed to generate synchronized video and audio within a single model. It brings together the core building blocks of modern video generation, with open weights and a focus on practical, local execution.<a target="_blank" rel="ugc" href="https://www.youtube.com/watch?v=8fWAJXZJbRA"><img src="https://img.youtube.com/vi/8fWAJXZJbRA/maxresdefault.jpg" alt="LTX-2 Open Source" /></a><h2 id="direct-use-license">Direct use license</h2>You can use the models - full, distilled, upscalers and any derivatives of the models - for purposes under the <a target="_blank" rel="ugc" href="https://huggingface.co/Lightricks/LTX-2/blob/main/LICENSE">license</a>.<h2 id="comfyui">ComfyUI</h2>We recommend you use the built-in LTXVideo nodes that can be found in the ComfyUI Manager. For manual installation information, please refer to our <a target="_blank" rel="ugc" href="https://docs.ltx.video/open-source-model/integration-tools/comfy-ui">documentation site</a>.<h2 id="general-tips:">General tips:</h2><ul><li>Width &amp; height settings must be divisible by 32. Frame count must be divisible by 8 + 1.</li><li>In case the resolution or number of frames are not divisible by 32 or 8 + 1, the input should be padded with -1 and then cropped to the desired resolution and number of frames.</li><li>For tips on writing effective prompts, please visit our <a target="_blank" rel="ugc" href="https://ltx.video/blog/how-to-prompt-for-ltx-2">Prompting guide</a></li></ul><h3 id="limitations">Limitations</h3><ul><li>This model is not intended or able to provide factual information.</li><li>As a statistical model this checkpoint might amplify existing societal biases.</li><li>The model may fail to generate videos that matches the prompts perfectly.</li><li>Prompt following is heavily influenced by the prompting-style.</li><li>The model may generate content that is inappropriate or offensive.</li><li>When generating audio without speech, the audio may be of lower quality.</li></ul><h2 id="huggingface-repository">HuggingFace Repository</h2><a target="_blank" rel="ugc" href="https://huggingface.co/Lightricks/LTX-2/tree/main">https://huggingface.co/Lightricks/LTX-2/tree/main</a>

Lightricks LTXV 2

The full model in nvfp4 quantization

lightricksLTXV2_ltx219bDevFp4.safetensors

ltx-2-19b-dev-fp4

Files with SHA256: 08a28245d8412962a8f6a1437c7db4b07bd5c2acdb2b84f96793a9b7c8550751