Mergenetic: a Simple Evolutionary Model Merging Library

Model merging allows combining the capabilities of existing models into a new onepost hoc, without additional training. This has made it increasingly popular thanks to its low cost and the availability of libraries that support merging on consumer GPUs. Recent work shows that pairing merging with evolutionary algorithms can boost performance, but no framework currently supports flexible experimentation with such strategies in language models. We introduce Mergenetic 1 , an opensource library for evolutionary model merging. Mergenetic enables easy composition of merging methods and evolutionary algorithms, while incorporating lightweight fitness estimators to reduce evaluation costs. We describe its design and demonstrate that Mergenetic produces competitive results across tasks and languages using modest hardware. ⋆ denotes equal contribution. 1 https://github.com/tommasomncttn/mergenetic Recent research has shown that combining model merging with evolutionary algorithms can achieve superior performance (Akiba et al., 2025; Mencattini et al., 2025) . However, this approach faces two key challenges: first, there is currently no library for experimenting with different evolutionary algorithms and merging methods; second, these methods typically require repeated evaluations on an evolutionary datasets to compute fitness functions, making them more computationally expensive than standard merging techniques. These limitations restrict access for the very user base that model merging was intended to empower. In this paper, we introduce Mergenetic, a simple library to easily perform evolutionary model merging. Built on top of MergeKit (Goddard et al., 2024) and the widely used evolutionary framework PyMoo (Blank and Deb, 2020), our library provides: 1. Comprehensive Algorithm Support. Mergenetic integrates 19 evolutionary algorithms and 6 merging strategies, enabling both single-and multi-objective optimization. This includes classical methods like genetic algorithms and state-of-the-art approaches such as NSGA-II (Deb et al., 2002a).