>
May, 14th, 2025: New distilled model 13B v0.9.7: Release a new 13B distilled model ltxv-13b-0.9.7-distilled Amazing for iterative work - generates HD videos in 10 seconds, with low-res preview after just 3 seconds (on H100)! Does not require classifier-free guidance and spatio-temporal guidance. Supports sampling with 8 (recommended), or less diffusion steps. Also released a LoRA version of the distilled model, ltxv-13b-0.9.7-distilled-lora128 Requires only 1GB of VRAM Can be used with the full 13B model for fast inference Release a new quantized distilled model ltxv-13b-0.9.7-distilled-fp8 for real-time generation (on H100) with even less VRAM