Real AI image generation in your browser
Pick a model, describe your image, hit Generate. Downloads a real Stable Diffusion model and runs it entirely in your browser via WebGPU. First run downloads the model — cached after that.
Requires Chrome/Edge 113+ with a decent GPU. Models range from 9MB (Tiny) to 4GB (styled models).
3 models: Tiny SD (9MB test), SD 1.5 (2GB, good quality), SD-Turbo (2.4GB, fastest). Styled models (4GB+) disabled — too large for browser memory.
Loading model...
Downloading model. This is a one-time download.