AI Model Globalizer

Manage AI models centrally and save storage space

AI Model Globalizer: Put an end to duplicate AI models on your SSD

Anyone who works with AI tools locally will be familiar with the problem: models quickly become huge. Checkpoints, LoRAs, ControlNets, VAEs, upscalers and GGUF files often end up in multiple copies across different programme folders. Each tool has its own structure, downloads models again or expects files to be in a specific location. The result: chaos on your hard drive – and eventually, even a large SSD can fill up surprisingly quickly.

That’s exactly why we’ve developed the AI Model Globalizer.

Links

Share

The Windows tool scans existing model folders, detects supported model files, and moves them into a central global model library. File links are then created at the original storage locations. For the respective AI software, everything continues to look exactly as it did before – while in reality, the file exists only once in a central location on the hard drive.

The major advantage: Existing workflows do not need to be rebuilt. The tools still find their models at the familiar paths, while storage space is saved and more structure is created in the background.

The new C# version is significantly faster and more convenient than the original batch script. It shows the scan progress live, first creates a plan, and only makes changes after explicit confirmation. Supported file types include .safetensors, .gguf, .ckpt, and .onnx.

The auto mode is especially practical: If the original path and the global library are located on the same NTFS drive, the tool uses hard links. Otherwise, symlinks can be used. In addition, there are check and repair functions to verify existing links or global model folders.

The AI Model Globalizer is available as a public GitHub project and is aimed at anyone working locally with AI models who wants to organize their model library in a cleaner, more space-saving, and more maintainable way. Disclaimer: Use at your own risk.

Have fun with it!

Archive