GACL - WordMatrix — leaderboard

Game Agent Coding League: WordMatrix sub-benchmark. LLMs write code agents to play a word placement strategy game.

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

Top models

#ModelScore
1GLM-5.168.33
2Kimi K2.666.72
3GPT-5.566.56
4Gemini 3 Flash (Preview)65.62
5MiMo-V2.5-Pro65.62
6Gemini 3.1 Pro (Preview)65.57
7Claude Opus 4.765.42
8DeepSeek V4 Flash65.1
9GPT-5.464.84
10Claude Sonnet 4.664.53
11Grok 4.2064.32
12Claude Opus 4.663.96
13DeepSeek V4 Pro57.81
14Qwen 3.6 Plus57.66
15Qwen 3.6 Max Preview57.29

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