Lightricks Releases LTX-2: Open-Source Audio-Video Generative Model and LoRA Trainer

Lightricks · github · 2026-08-12

Lightricks, a well-known image and video processing company, has officially released the Python inference and LoRA trainer package for LTX-2 on GitHub.

As an audio-video generative model, this release allows developers to deploy LTX-2 locally for inference and train custom LoRAs for specific use cases, further lowering the barrier for audio-visual content creation.

Original post →

More from Multimodal

Multimodal channel →