Model Management
Download, organize, and cache Stable Diffusion models
Adding Models
ALICE automatically detects models placed in the models directory.
Supported Formats
- Safetensors (recommended)
- PyTorch (.pt, .pth, .bin)
- Diffusers format (directories)
Model Sources
CivitAI
Community models with previews and ratings. Download and place in models directory.
HuggingFace
Official Stable Diffusion models and variants.
Auto-Detection
ALICE automatically identifies:
- SD 1.5 models (512x512 native)
- SDXL models (1024x1024 native)
- FLUX models (flexible resolution)
- Turbo models (1-4 steps)
Model Caching
Configure cache behavior:
cache_size- Number of models in memoryauto_unload- Free GPU memory when idle- Models load on first use
- Cached models generate instantly
Best Practices
- Use safetensors format for safety
- Organize models in subdirectories
- Keep model count reasonable for your VRAM
- Use descriptive filenames
Next Steps
- Supported Models - Compatibility list
- Performance - Optimization tips
- Installation - Setup guide