GPT-3.5 Standard vs GPT-3.5 Aggressive | AI vs AI Chess Play | Code available
🔗 GitHub: https://github.com/konradcinkusz/llm-...
I pit two AI agents, both configured with an aggressive personality, against each other on the chessboard. This match is powered by my open‑source Chess‑Agent System with LLM, which uses a multi‑agent architecture to let large language model agents play full games of chess.
This project is more than a demo – it’s a flexible framework for AI‑driven chess simulations. Key features include:
Multi‑Agent System – two LLM‑powered agents battle head‑to‑head, making moves autonomously
Configurable Agents – choose different OpenAI models (such as GPT‑4o) and personalities to see how play styles change
Interactive UI & Move History – built with Streamlit, it shows real‑time boards, detailed move tracking and auto‑play/step‑through controls
Game Analysis – material balance, position evaluation and logging for deeper insights
Headless Simulations – run matches via a script without launching the UI for quick experiments or batch runs
If this match sparks your curiosity, explore the full source code and instructions here:
🔗 GitHub: https://github.com/konradcinkusz/llm-...
The repository includes installation steps (clone the repo, create a virtual environment, install requirements and set your OpenAI API key), usage instructions (configure agents and control the game through the sidebar) and details on running headless simulations. It’s licensed under MIT and contributions are welcome—feel free to fork, add features, write tests and open a pull request.
Enjoy the match and let me know what configurations you’d like to see next!