Grand Unified Model Merger V2
A lightweight Windows-friendly Python GUI for merging Stable Diffusion checkpoint .safetensors models.
V2 focuses on safer base checkpoint merging with a cleaner interface, easier blend controls, and preflight checks before you commit to a merge.
Features
Weight Sum, Add Difference, and TAME merge modes
Locked-weight auto-balance for easy blend ratios
Manual blend weights that do not fight you while typing
Preflight check before merging
Overwrite confirmation before replacing an existing output file
Recent base model and output folder memory
Preset save/load
Dark V2 interface with custom header art
Blend Controls
Unlocked rows auto-balance automatically.
If you type a weight and press Enter, that row locks. The remaining unlocked rows split the leftover weight.
Example:
Model A locked at 0.6
Model B auto-balances to 0.2
Model C auto-balances to 0.2
Unlock a row to return it to auto-balance.
Important
LoRA checkpoint injection is disabled in V2. This version is focused on base checkpoint merging because LoRA baking was unreliable in earlier testing.
Install
Extract the zip.
Install Python 3.10+.
Run install_requirements.bat.
Run run_model_merger_v2.bat.
Requirements
Python 3.10+
torch
safetensors
If torch does not install from the included batch file, install the PyTorch build for your CUDA/CPU setup from the official PyTorch site.
Description
Grand Unified Model Merger V2
A lightweight Windows-friendly Python GUI for merging Stable Diffusion checkpoint .safetensors models.
V2 focuses on safer base checkpoint merging with a cleaner interface, easier blend controls, and preflight checks before you commit to a merge.
Features
Weight Sum, Add Difference, and TAME merge modes
Locked-weight auto-balance for easy blend ratios
Manual blend weights that do not fight you while typing
Preflight check before merging
Overwrite confirmation before replacing an existing output file
Recent base model and output folder memory
Preset save/load
Dark V2 interface with custom header art
Blend Controls
Unlocked rows auto-balance automatically.
If you type a weight and press Enter, that row locks. The remaining unlocked rows split the leftover weight.
Example:
Model A locked at 0.6
Model B auto-balances to 0.2
Model C auto-balances to 0.2
Unlock a row to return it to auto-balance.
Important
LoRA checkpoint injection is disabled in V2. This version is focused on base checkpoint merging because LoRA baking was unreliable in earlier testing.
Install
Extract the zip.
Install Python 3.10+.
Run install_requirements.bat.
Run run_model_merger_v2.bat.
Requirements
Python 3.10+
torch
safetensors
If torch does not install from the included batch file, install the PyTorch build for your CUDA/CPU setup from the official PyTorch site.