LTX-2.5 Open-Weights Video Model Targets Local NVIDIA RTX Inference · cho.sh
LTX-2.5 Open-Weights Video Model Targets Local NVIDIA RTX Inference
AI video model releases
LTX-2.5 Open-Weights Video Model Targets Local NVIDIA RTX Inference
LTX-2.5 is an open-weights video world model for text-to-video and image-to-video generation that runs locally on NVIDIA RTX GPUs and NVIDIA DGX Spark.
LTX reports that 2x NVIDIA GB200 hardware generates a 10-second clip in 6.8 seconds; its API variant takes 23.7 seconds.
The model uses a Gemma 4 vision-language backbone, a decoder intended to reduce artifacts in high-motion footage, and native multishot generation to retain character identity across shots.
Developers and studios can fine-tune LTX-2.5 with custom LoRA adapters in ComfyUI, according to the article.
cho.sh