...It installs an isolated inference engine for you — no terminal, no CUDA wrestling, no Python environment. It auto-detects your GPU and tunes offloading/quantization to fit your VRAM: 8GB runs SDXL smoothly. Run seven model families locally (SDXL, SD 1.5, Z-Image, FLUX, Chroma, Qwen-Image, Anima), load your own .safetensors from Civitai, and keep every prompt and image on your machine. Optional cloud generation for bigger models — no account needed. Windows & macOS. Fully localized in English and Chinese.