chess5.ai

Human vs LLM · Five Games

Pit yourself against GPT, Claude, Gemini, Grok, Muse Spark, Mistral, DeepSeek, Kimi, Qwen, GLM, or MiniMax across five classic boards.

0 matches played

Enter Arena Configure Keys

Five games

Chess
Standard international chess.
Go
Captures and simple ko; ends after two passes.
Xiangqi
Chinese chess on a 9×10 board.
Gomoku
15×15 free-style, five in a row wins.
Othello
8×8 Reversi. Flip discs; most discs wins.

How it works

Human vs model, or model vs model — with spectating and auto-play built in.
The model chooses each move entirely on its own.
The app parses the move and decides the result by the game rules.
Three illegal model outputs in one match also count as a loss.
If you leave the play page, the match ends with no winner.

About

This project is a fun experiment that tests each model's own board-game strength — how smart it really is. Difficulty is adjustable, so it's also nice for beginners and kids learning board games against models. The project is open source — explore the source code.