Mergoo provides tooling to combine multiple specialized LLM experts (e.g., domain-specific or fine-tuned adapters) into a single model. It supports Mixture-of-Experts (MoE) and Mixture-of-Adapters (MoA) merging strategies, as well as layer-wise merging for fine-grained control. The library integrates with HuggingFace Transformers and PEFT for training, allowing users to either train only the routing layers or perform full fine-tuning of the merged model on standard hardware including CPU, MPS, and GPU.