Open Generative AI
lastmod 2026-05-07created 2026-05-073 edits
A studio for generating images/videos with 200+ AI models
About
A free, open-source alternative to Higgsfield AI, Freepik, Krea AI, and Openart AI
Supports text→image, image→image, text→video, image→video, and lip-sync generation without content filters, through 200+ models including Flux, Midjourney, Kling, Sora, and Veo
- Desktop app installers for macOS (Apple Silicon/Intel), Windows, and Linux (Electron); the web-hosted version can be used at dev.muapi.ai without installation
- Next.js monorepo architecture, a packages/studio shared component library, and two-step pattern (Submit → Poll) communication through the Muapi.ai API gateway
- MIT license; self-hosting and customization possible
Five studios
- Image Studio: dual mode — 50+ t2i and 55+ i2i models, up to 14 reference images
- Video Studio: 40+ t2v and 60+ i2v models
- Lip Sync Studio: 9 models — portrait + audio or video + audio
- Cinema Studio: pro camera controls — lens, focal length, aperture
- Workflow Studio: node-based multi-step pipeline builder
Two local inference engines supported (desktop app only)
- sd.cpp (bundled, Apple Silicon Metal/CUDA/Vulkan/ROCm — SD 1.5, SDXL, Z-Image)
- Wan2GP (user-run Gradio server — Flux, Qwen-Image, Wan 2.2 T2V/I2V, Hunyuan, LTX video)
Recently added models
- Nano Banana 2 (based on Google Gemini 3.1 Flash, up to 4K, 14 reference images)
- Seedream 5.0 (ByteDance)
- MiniMax Image 01
- Seedance 2.0 (T2V/I2V/Extend)
- Grok Imagine (xAI, T2V/I2V)