编排:Gemini-3-5-pro

这不是客观事实。 本页记录本站的梯队、星级、六维和情绪等主观判断，与模型事实页物理分离，仅编排员和管理员可修改。

<syntaxhighlight lang="yaml"> id: gemini-3-5-pro tier: T2 stars: 4 composite: 84 stats:

 code: 52
 reasoning: 74
 context: 76
 speed: 55
 multimodal: 90
 value: 48

sentiment:

 positive: 20
 mixed: 25
 negative: 55

radar: - axis: 长程任务

 value: 72

- axis: 编程工程

 value: 52

- axis: 抽象推理

 value: 74

- axis: 上下文利用

 value: 76

- axis: 中文能力

 value: 50

- axis: 响应速度

 value: 55

- axis: 稳定性

 value: 38

- axis: 指令遵循

 value: 56

- axis: 易用性

 value: 42

- axis: 性价比

 value: 48

danmaku: - text: new checkpoint is deadass insane

 platform: reddit
 main: true

- text: The END · 942 票哀嚎

 platform: reddit
 main: true

- text: 3.5 Pro High 5 分钟完成 Fable 5 半小时任务

 platform: hn
 main: true

- text: 3.5 Pro 相关 HN 主流报道合计 ≤18 分

 platform: hn
 main: true

- text: 又一次跳票，硅谷最尴尬 AI 模型

 platform: zhihu
 main: true

- text: 御三家要换人了，grok 可上位

 platform: linuxdo
 main: true

- text: fundamentals are shaky and old

 platform: reddit
 main: false

- text: make stuff you didn't ask for

 platform: reddit
 main: false

- text: It didn't beat opus 4.8 on a single benchmark

 platform: reddit
 main: false

- text: Gemini 3.5 deleted 28,745 lines, broke production

 platform: hn
 main: false

- text: Gemini models increasingly stucking in thinking loop

 platform: hn
 main: false

- text: consistently the most frustrating model for development

 platform: hn
 main: false

- text: Google has become the laughing stock of frontier AI

 platform: x
 main: false

- text: best UI/design taste yet

 platform: x
 main: false

- text: 发布一周后被削智，Gemini 周期律

 platform: zhihu
 main: false

- text: 已被悄悄取消？转推 Gemini 4

 platform: linuxdo
 main: false

</syntaxhighlight>