Open Generative AI GitHub Repo: A Self-Hosting Teardown (2026)

Description and use

Open Generative AI offers a MIT‑licensed UI that aggregates hundreds of cloud models and two local engines (sd.cpp for images, Wan2GP for video). It is good for prototyping but needs a Muapi key for cloud usage and powerful hardware for local inference.

Problem

Clients want a free, self‑hosted generative‑AI UI but the repo’s open‑source code hides that many models are still cloud‑based and local inference has high hardware costs.

Outcome

The repo offers a comprehensive UI but requires a Muapi key for most models and significant hardware for local inference. It is suitable for prototyping and side‑by‑side model comparison, but not ideal for fully offline production workflows.

Workflow

• Clone repo with submodules
• Run npm run setup to install dependencies and build packages
• Choose npm run electron:dev for desktop or npm run dev for web
• Configure Muapi key in settings for cloud models
• Install sd.cpp via Settings for image inference
• Install Wan2GP on a CUDA/ROCm GPU and configure MCP URL for video inference

Evidence

If you search "open generative ai github" in 2026, the repo you land on is Anil‑matcha/Open‑Generative‑AI: an MIT‑licensed Next.js and Electron studio for image, video, and lip sync generation, at roughly 29.7k stars and 5.4k forks as of October 6.

Source

Cloud generation goes through the Muapi.ai API with your own key, and truly local inference only exists in the desktop app, through two engines with real hardware requirements.

Source

The README is clear that most people should grab a prebuilt installer. If you want to hack on it, this is the path, and I checked each script name against package.json: ... npm run setup ... npm run electron:dev ... npm run dev ... npm run build && npm run start

Source

sd.cpp (bundled). Built on stable‑diffusion.cpp. You install it with one click under Settings, then Local Models. It handles image models only: Dreamshaper 8, Realistic Vision 5.1, and Anything v5 (SD 1.5, about 2.1 GB each), SDXL Base 1.0 (6.9 GB), and Z‑Image Turbo and Base, which also need a 2.4 GB Qwen3‑4B text encoder and a 335 MB FLUX VAE.

Source

Wan2GP (bring your own). This is where video lives. The app doesn’t bundle Python or weights. You install Wan2GP yourself on a CUDA or ROCm machine, and the app either runs wgp.py from a local folder or talks to Wan2GP’s MCP server on another box.

Source