Genmo
Free tierOpen-source video world models that understand the physical world like never before
Free tier available·All audiences·Powered by Genmo (Mochi 1)·Open source
Key strengths
State-of-the-art open-source text-to-video model (Mochi 1)Fully self-hostable and locally runnableComfyUI integration for no-code customizationInteractive playground for prompt-based video generationResearch-grade model with physical world understanding
Free tier + paid plans
US
Self-hostable
No ratings yet
Developer Setup for Mochi 1
Quick Start
# 1. Clone the repository
git clone https://github.com/genmoai/mochi
# 2. Install Python dependencies
pip install -e .
# 3. Generate your first video via CLI
python demos/cli.py
Key Resources
- GitHub:
github.com/genmoai/mochi— Full source code, model weights, and contribution guide. - HuggingFace: Model weights and tokenizer artifacts hosted for easy download.
- ComfyUI Integration: Drop-in node support for visual, no-code pipeline building with Mochi 1.
Notes for Developers
- The model is open-source and licensed for local use and customization.
- GPU performance engineering is an active focus area (per hiring), so hardware-accelerated inference paths are being developed.
- The CLI (
demos/cli.py) accepts text prompts and outputs video files directly. - Contributions to the repo are welcomed — the project is community-driven.
