Open text-to-video from Genmo. Run the weights yourself or use the hosted app for coherent motion.
Mochi, from Genmo, was one of the first open text-to-video models to rival closed rivals on motion quality. The open weights let you self-host; the hosted app gives you prompts-to-clip with surprisingly coherent movement and decent prompt following.
Who it's for: Creators and builders who want video generation without a paywall, or who need to self-host for control and privacy.
Describe a scene and get a few seconds of coherent motion.
Run Mochi-1 locally if you have the GPU — no usage caps.
Better limb and physics handling than many early open models.
Animate a still image with directed camera and motion.
Mochi is the open video generator to watch in 2026. Use the hosted app for quick experiments; self-host the weights when you need volume or privacy. Pair with Runway or Kling for production finish.