Open Generative AI
Desktop (Electron) and web app that aggregates 200+ image and video models (Flux, Sora, Veo, Kling, Midjourney-like, Runway, Hunyuan, Wan 2.2) behind a glassmorphic tabbed UI. Hybrid: web version is 100% cloud via muapi.ai (API key), desktop version is self-hosted with bundled sd.cpp for local models + optional Wan2GP BYO server for video. MIT licensed, lip sync built in (9 models).
Strengths
- Ergonomic tabbed UI: 200+ image/video models accessible without learning a node-based pipeline
- Desktop version bundles sd.cpp for 100% local execution (prompts and assets never leave the machine)
- Built-in lip sync (9 models), a rare feature combined with an image+video studio
- MIT licensed, fully hackable app, desktop install via DMG/EXE without Node.js
Limitations
- Explicit "uncensored / no content filters" positioning, no articulated ethical framework, use with full awareness of legal risks (deepfakes, GDPR, image rights)
- Web version is 100% cloud-dependent on muapi.ai, prompts and assets transit through a commercial third party
- Ambiguous local/cloud hybrid: only desktop + sd.cpp is truly self-hosted; everything else is API proxy
- Less fine pipeline control than ComfyUI (no node graph, closed presets)
Best for
- Quick personal benchmark of several generative models before deciding which to integrate into a ComfyUI pipeline or production
- Solo creators who want a packaged image+video+lip sync studio without building the stack themselves