ComfyUI Model Randomizer

A custom node for ComfyUI that randomly selects a checkpoint from a configurable list of models. Each model entry has its own CFG range, probability weight, and enable/disable toggle.

Replaces the common chain of Load Checkpoint → Context Big → Fast Muter → Random Unmuter with a single, self-contained node.

Features

• Weighted random selection — assign probability weights to favor certain models
• Per-model CFG range — set a min/max CFG for each entry; a random value is picked within the range
• Enable/disable toggle — temporarily exclude models without removing them
• Seed control — use a fixed seed for reproducible selections, or 0 for fully random
• Checkpoint caching — skips reloading when the same model is selected consecutively
• Drag-to-reorder — move entries up/down with built-in buttons
• Add/remove entries — dynamically grow or shrink the model list

Outputs

| Output | Description | |---|---| | MODEL | Diffusion model from the selected checkpoint | | CLIP | CLIP model from the selected checkpoint | | VAE | VAE model from the selected checkpoint | | MODEL_NAME | Filename of the selected checkpoint | | CFG | Random CFG value within the selected model's range |

Installation

Clone into the custom_nodes directory:

cd ComfyUI/custom_nodes
git clone https://github.com/Mervent/comfyui-model-randomizer.git

Restart ComfyUI. The node appears under loaders → Model Randomizer.

Usage

1. Add the Model Randomizer node to your workflow
2. Click Add Model to create model entries
3. For each entry, select a checkpoint and configure:
   • CFG Min / Max — the range for random CFG generation
   • Weight — relative probability (higher = more likely to be picked)
   • Enabled — toggle to include/exclude from selection
4. Connect the outputs to the rest of your pipeline

Set seed to a non-zero value for reproducible results across runs, or leave at 0 for a fresh random pick every time.

License

MIT