Carrick AI GameBench Benchmarking AI coding through games用游戏评测 AI 编程能力
Versioned by design持续迭代,版本可追溯

The benchmark evolves. Scores stay comparable.测试集会迭代,成绩仍然可比较。

Games, tests, and scoring rules can improve over time. Every result names its exact release, and rankings never mix incompatible versions. 游戏任务、测试方法和评分规则都会继续改进。每条成绩都会标明对应版本,不同版本不会混在一起排名。

Current release当前版本

v0.4.0

The public games and rankings use this ruleset by default.公开游戏与评测排行默认使用这一版规则。

Historical release archive · 5 retained历史版本档案 · 已保留 5 个版本

Earlier releases remain available for reproducibility and old result links. Most visitors do not need them.历史版本仅用于复现旧成绩和保持旧链接有效,普通用户通常不需要查看。