LiveCodeBench

What it tests: Competitive-programming problems (LeetCode/AtCoder/Codeforces) collected continuously after model training cutoffs, scored by pass rate with zero contamination by design.

考什么:持续从 LeetCode/AtCoder/Codeforces 采集的新竞赛题,题目发布晚于模型训练截止日,从机制上杜绝刷题污染,按通过率计分。

Why it counts: The anti-contamination gold standard for coding — a high score here is much harder to fake than on static benchmarks.

为什么算数:代码领域防污染的标杆——题库月月更新,在这里拿高分比在静态 benchmark 上难作弊得多。