Mochi 1

Mochi 1 is a 10-billion-parameter open-source text-to-video generation model released by Genmo. It turns text prompts into short video clips, and because its weights are published, it suits users who want to run the model on their own GPUs rather than depend on a hosted service. Its primary audience is researchers and developers who work with video generation at the code level.

Its biggest strengths are licensing and control. Released under Apache 2.0, it carries no restrictions on commercial use, and self-hosting keeps your data and pipeline in your own hands. Outputs leave no watermark, so they drop straight into editing and compositing. In our evaluation its value-for-money score of 90 stands out, reflecting the nature of a free open model you download and run. It also exposes an API, leaving room for automation.

The limits center on the runtime environment and output specs. The recommended setup is an H100-class GPU with roughly 60GB of VRAM, which is hard to meet on typical personal hardware. Each clip tops out at 5 seconds and 480p, making it closer to a prototyping tool than a finished-content engine.

In short, it fits users with their own GPU infrastructure who want an open model that is also free to use commercially. If you instead need high-resolution, longer footage or the convenience of a tool that works without setup, a commercial cloud video service is the better place to look.