SDXL Node Merger - A new method for merging models. OPEN SOURCE

A developer known as anonimgeronimo has launched SDXL Node Merger, a free and open-source tool that provides a visual, node-based interface for creating and testing merged Stable Diffusion XL models. Unlike command-line tools or basic UIs, it allows users to drag, drop, and connect nodes with animated Bezier curves to design everything from simple two-model blends to intricate multi-model chains. The tool supports 11 different merge algorithms—including Weighted Sum, Add Difference, TIES, DARE, and SLERP—all with Merge Block Weighted (MBW) support for fine-grained control. A key productivity feature is batch processing: users can set up multiple merge configurations with different Save nodes and execute them all at once, returning later to find several finished models.

Built with the latest CUDA 12.x and PyTorch, the tool is ready for modern hardware like the RTX 50-series and includes a low VRAM mode that streams tensors sequentially, enabling merges on GPUs with as little as 4GB of memory. It comes with four visual themes and a simple setup process that handles the virtual environment and dependencies automatically, launching directly in a browser. This approach transforms model merging from a tedious, manual process into a visually intuitive and highly efficient workflow, significantly speeding up experimentation for AI artists and researchers.