ltx2.site

LTX-2, accessible via ltx2.site, is a groundbreaking open-source multimodal AI model developed by Lightricks. It sets a new standard for synchronized audio-video generation, producing coherent video clips with matching audio in a single, unified process. Unlike traditional tools that generate silent footage requiring separate audio dubbing, LTX-2 natively creates videos up to 20 seconds long with perfectly synced dialogue, sound effects, and background music. It is engineered for high-end output, supporting up to 4K resolution and high frame rates, making it suitable for professional-grade content. The model is completely free and designed for local deployment on consumer NVIDIA GPUs, offering creators full control and privacy. With native support for ComfyUI, it integrates seamlessly into advanced AI workflows. ltx2.site is the hub for developers, indie filmmakers, content creators, and researchers seeking a powerful, open, and self-hosted solution for generating high-fidelity audio-visual narratives without subscriptions or cloud dependencies.