MochiUnclaimed

Genmo is a research lab that builds open video generation models. Its main product is Mochi 1, an open-source text-to-video model that turns written descriptions into short video clips.

The page describes Mochi 1 as a model for turning written concepts into visual stories. Users can run it locally, customise it for their own needs, or contribute to its development. The code is published in a GitHub repository, and the model weights are available on Hugging Face. The page also lists ComfyUI as a way to run and customise Mochi.

For people who want to try the model without setting anything up, Genmo offers an interactive playground where they can test Mochi's features and capabilities. The homepage includes a random prompt generator with example descriptions, such as a slow-motion shot of a glass shattering or a time-lapse of a chalk mural being painted.

To run Mochi from the command line, the page gives three steps: clone the repository, install the dependencies with pip, and run the demo script to generate a first video.

Genmo describes its wider goal as building open, state-of-the-art models for video generation, with the aim of developing world models that understand the physical world. The company is hiring for research and engineering roles, including research scientists working on diffusion and post-training, a new-grad research engineer and a GPU performance engineer.

The page does not state any prices for Mochi or the playground.