WHAT NEW IN X10.29 THU

X 上的 AI 合集档案 · 22 个合集 · 60,636 条

minimax-m3.1POST

用户反馈:Space Bunny(M3.1 Flash 匿名版)盲测中与 Opus 5 不相上下

评论者认为盲测揭晓是最好的评价方式:在不知道品牌的情况下,Space Bunny 的输出能与 Opus 5 抗衡,比任何跑分表都更有说服力。

CloudDog
@CloudDogx8 天前收录原推发布于 2026.10.20
CloudDog @CloudDogx原推发布于 2026.10.20
在 X 上打开原推

@k2sbhai The blind-test reveal is the best review a model can get — you judged it purely on output, no brand halo. Space Bunny holding its own against Opus 5 without you knowing what it was says more than any benchmark table. Stealth evals should be the default.

x.com/clouddogx/status/2112230110346584359采集于 2026.10.20 20:42