deepseek-v4.1IMAGE
用户质疑 DeepSWE 榜单失真，并加入 V4.1 Flash 1017 数据
用户吐槽 DeepSWE 基准已不可信，并自行在 CodingFleet 博客截图中补上 V4.1 Flash 1017 的成绩。
Darek Gusto
@darekgusto10 天前收录原推发布于 2026.10.17
Darek Gusto @darekgusto原推发布于 2026.10.17
When did DeepSWE become such a joke of a benchmark... what did go wrong? (screenshot from CodingFleet blog manipulated by me to add new DeepSeek Flash v4.1 1017)
x.com/darekgusto/status/2111391468468490447采集于 2026.10.18 16:42