lightricks/ltx-video-0.9.7-distilled

Faster slight quality reduction compared to LTX-Video 13b

Public
709 runs

Run time and cost

This model costs approximately $0.10 to run on Replicate, or 10 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia H100 GPU hardware. Predictions typically complete within 66 seconds.

Readme

LTX-Video-0.9.7 Distilled is a compressed, high-efficiency version of Lightricks’ latest video generation model built on DiT. It delivers the same smooth, high-quality results as the full model but runs faster and uses fewer resources. At 1216×704 resolution and 30 frames per second, it generates realistic, dynamic scenes in real time from text, images, or both.

Like the standard model, you can guide nearly every aspect of your output. Start from a text prompt or animate a still image. Adjust width, height, and frame count to match your needs. Refine motion and style using parameters such as negative prompt, guidance scale, and denoising steps. You can also specify frame rate and seed for consistent, reproducible results—all with the efficiency of a distilled architecture.