今天新增gemini-4IMAGE
Gemini 4 Carbon 基准成绩被质疑 benchmaxxing
用户称 Gemini 4 Carbon 的基准数据夸张得离谱,但怀疑是针对榜单刷分,要等实测才能定论。
Leon Lin
@LexnLin5 小时前收录原推发布于 2026.10.29

Leon Lin @LexnLin原推发布于 2026.10.29
Woah! These benchmark results from Gemini 4 Carbon look insane! But it's a Gemini model and I smell the benchmaxxing. Anyways testing it when it's out is the only way to find out whether it's good or not
这条是引用着另一条写的:「Google DeepMind 官宣 Gemini 4 Carbon 为下一代前沿模型」
x.com/lexnlin/status/2115493815054180826采集于 2026.10.29 10:01