LTX-2.5
LTX-2.5 is an open-weight video-audio world model from Lightricks that generates multiple connected scenes in a single pass, maintaining consistent character identity, environment, and voice across cuts. It supports text-to-video, image-to-video, video-to-video, audio-to-video, and text-to-audio-video, featuring a new video decoder for sharper faces and fewer artifacts, a custom Gemma 4 12B text encoder, and a prompt enhancer; it is self-hostable with no per-generation billing and available on Hugging Face under a commercial license.