MOSS-Image-Base
Text-to-image demo for OpenMOSS-Team/MOSS-Image-Base. Prompts can be natural language, tags, or a JSON layout. Defaults follow the model card: 1024×1024, 50 steps, guidance 3, shift 5.
The Gradio app is in this repo (app.py). It needs a GPU, and this account cannot attach ZeroGPU yet. A community GPU grant is requested on the Community tab. After it is approved, the Space SDK switches to Gradio and the demo starts serving images.
Faster path: subscribe to Hugging Face PRO, then this Space can be recreated on ZeroGPU immediately.