Mochi 1

Mochi 1 is a video generation model from Genmo, released as a research preview. It takes a text prompt and produces video, with the stated focus on motion quality, prompt adherence, and coherent human movement and expression rather than the stiff results that earlier video models often produced. There are two ways to use it. The online Playground lets you type a prompt and see a result quickly, and because the model is open source you can also download the weights and run or customize it in your own environment. A free trial is currently available, and the page does not publish detailed pricing.

- Generate video from a text prompt describing characters, scenes, and actions

- Experiment in the Playground without local setup

- Download the open weights for local deployment and modification

- Produce clips aimed at film, game, and educational use cases

- Iterate on a prompt until the motion and framing look right

- Researchers and developers studying video generation models

- Filmmakers and game developers who need generated footage

- Educators building reenactments or visual explanations

- Hobbyists who want to experiment with open video models locally

This is a research preview, which means output quality, clip length, and stability will vary and may change between versions, so do not build a production pipeline on a single release. Detailed pricing is not listed, so treat commercial plans as unconfirmed until the provider publishes them. Running the weights locally requires appropriate hardware and some setup, including storage for the model files and time to learn the workflow. Generated footage also still needs review for quality, accuracy, and rights before it goes into a released project, and prompts should avoid asking for real people or protected characters.