Apple Silicon LTX-2.5 inference tool for AI video generation
The project adds LTX‑2.5 compatibility to the existing ltx‑2‑mlx codebase, enabling text‑to‑video, image‑to‑video, and audio‑conditioned video generation on Apple Silicon. It offers low‑memory streaming, model quantization (int8/int4) and various up‑sampling options, making high‑resolution generation feasible on consumer Macs. Targeted at creators and researchers who need on‑device, fast video synthesis without cloud resources. Compared to generic ML inference scripts, it provides a ready‑to‑run CLI with many production‑grade features.
View on GitHub →TommyKammy/ltx-2.5-mlx-inference