GACL - 2x8 MiniChess — leaderboard

Game Agent Coding League: 2x8 MiniChess sub-benchmark. LLMs write code agents to play a simplified chess variant on a 2x8 board.

Metric: Normalized Score (0-100). Source: gameagentcodingleague.com. Status: saturation imminent. 21 models tracked.

Top models

#ModelScore
1GPT-5.569.06
2GPT-5.467.4
3Claude Opus 4.660.21
4MiMo-V2.5-Pro52.08
5Gemini 3.1 Pro (Preview)50.83
6GLM-5.150.47
7GPT-5.4 Mini49.43
8Kimi K2.648.12
9Qwen 3.6 Max Preview45.47
10Claude Sonnet 4.643.8
11DeepSeek V4 Pro43.8
12Claude Opus 4.743.18
13Qwen 3.6 Plus40.42
14GPT-OSS-120B38.85
15Qwen 3.6 35B A3B38.12

Interactive version: theaggregate.ai/benchmark?slug=gacl-2x8-minichess · How the rankings work · Data refreshed daily, snapshot 2026-07-22.