# DeepSWE发布：揭示模型真实编码能力差距

- 来源：SemiAnalysis (@SemiAnalysis_)
- 发布时间：2026-05-28 02:55
- AIHOT 分数：36
- AIHOT 链接：https://aihot.virxact.com/items/cmpoftgwf05rkslv4bgyndo3w
- 原文链接：https://x.com/SemiAnalysis_/status/2059710098650681846

## AI 摘要

这里有一个非常重要的教训，但你们中的一些人还没准备好进行这场对话。

## 正文

there's a really important lesson here， but some of yall aren't ready for that conversation yet

### 引用推文

> Serena Ge (Datacurve)：Today we're releasing DeepSWE, a new standard for agentic coding benchmarks. On public leaderboards, top models often look relatively close in capability. DeepS...