DeepSeek v4.1 Flash 在 ErdosBench 上与 Gemini 4 Argon 打平

ErdosBench 226 道开放数学题上，DeepSeek v4.1 Flash 1017 排第 6，与 Gemini 4 Argon、Muse Wonder 同分但成本低得多。

Przemek Chojecki | PC

@prz_chojecki1 天前收录原推发布于 2026.10.27

DeepSeek v4.1 Flash 在 ErdosBench 上与 Gemini 4 Argon 打平

Przemek Chojecki | PC @prz_chojecki原推发布于 2026.10.27

DeepSeek v4.1 Flash 1017 scored 6th on our ErdosBench - a collection of 226 open problems inspired by Erdos problems. It tied with Gemini 4 Argon and Muse Wonder at a fraction of a cost.

x.com/prz_chojecki/status/2114749560983892341采集于 2026.10.27 09:53

在 X 上打开原推