Codebase-scale open source evolutionary program synthesis inspired by AlphaEvolve
OpenEvolve is an open-source framework for evolutionary program synthesis, inspired by the AlphaEvolve project by DeepMind. It leverages evolutionary algorithms and large language models to automatically generate, mutate, and evaluate code, enabling the discovery of novel solutions to programming tasks.
- Codebase-Scale Optimization Contrary to previous implementations, OpenEvolve allows for codebase-scale optimization, as opposed to just one function.
- Evolutionary Search: Uses evolutionary strategies to explore the program space.
- LLM Integration: Incorporates large language models for code mutation and synthesis.
- Asynchronous Evaluation: Runs and evaluates generated code asynchronously in a sandboxed environment.
- Extensible: Easily add new tasks and evaluation metrics.
OpenEvolve requires Docker to run (install instructions). For the remaining requirements, please follow the steps:
# Clone the repository
git clone https://github.com/ryanrudes/openevolve.git
cd openevolve
# (Optional) Create a virtual environment
python3 -m venv venv
source venv/bin/activate
# Install dependencies
pip install -e .If you use this project, please cite it as such:
@software{openevolve,
title = {OpenEvolve: Codebase-scale evolutionary algorithms for code optimization, powered by LLMs},
author = {Ryan Rudes, James Hou, and Aditya Mehta},
year = {2025},
howpublished = {\url{https://github.com/ryanrudes/openevolve}},
note = {Open-source evolutionary program synthesis framework. Based on original code licensed under Apache License 2.0 by the FunSearch project.}
}
This project is licensed under the terms of the MIT license. See LICENSE.txt for details.
OpenEvolve: Codebase-scale evolutionary algorithms for code optimization, powered by LLMs.