跳到主要内容
严格题归零 严重度 8/10 2026-W18

Claude Sonnet 4.6 综合严格题归零

Claude Sonnet 4.6 Run #87

分数对比

维度 上期 本期 变化
主榜 (v5) 64.5 66.2 +1.7
代码执行 (v5) 88.0 92.8 +4.8
知识综合 (v5) 55.2 50.8 -4.4
材料约束 (v5) 74.9 87.1 +12.2
性价比 24.2 25.1 +0.9
稳定性 38.5 35.7 -2.8
可用性 100.0 99.0 -1

受影响维度

execution

丢分题目 Top 1

#1 SQL:疑似重复支付识别 100 0
模型原始回复(截取)
```sql
SELECT 
    p1.id AS first_id,
    p2.id AS second_id,
    p1.user_id,
    p1.merchant_id,
    p1.amount
FROM payments p1
JOIN payments p2 ON 
    p1.user_id = p2.user_id 
    AND p1.merchant_id = p2.merchant_id 
    AND p1.amount = p2.amount
    AND p1.id < p2.id
WHERE 
    p1.status = 'paid' 
    AND p2.status = 'paid'
    AND ABS(EXTRACT(EPOCH FROM (p2.created_at - p1.created_at))) <= 120
ORDER BY first_id, second_id;
```
Run #87 · 公式 v7 · 判分 v6 · 题库 v6 · 2026-04-27 04:18 SGT
查看 Claude Sonnet 4.6 完整档案