LTX 2.3

This site is a workspace for generating video with the LTX model, named LTX 2.5 on the page although the product name is LTX 2.3. You can type a scene, drop in a photo or upload an audio track, and the clip renders on cloud GPUs. The page claims 4K output at up to 50 frames per second, clips of up to 20 seconds, camera motion you choose and sound that arrives locked to the picture, with a faster tier for iteration and previews.

Text to Video returns a finished scene from a descriptive prompt, including lens, light, pacing and mood. Image to Video keeps the subject, palette and framing of a still while adding motion, and Audio to Video animates to a soundtrack you supply. The release notes on the page describe a new video decoder, a text encoder that follows complex prompts, multi-shot sequences that keep the same character across cuts and automatic duration. An AI image generator is also available.

Generation spends credits, shown per run in the interface, and a pricing page lists the credit options. The page includes a compliance note about model routes and moderation, and it shows sample prompts such as a citadel scene and a village rider.