编排:Gemini-3-1-pro

主观编排gemini-3-1-procuration/models/gemini-3-1-pro.yml

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-3-1-pro tier: T0 stars: 6 composite: 94.6 stats:

 code: 74
 reasoning: 92
 context: 86
 speed: 68
 multimodal: 90
 value: 80

sentiment:

 positive: 55
 mixed: 10
 negative: 35

radar: - axis: 长程任务

 value: 62

- axis: 编程工程

 value: 74

- axis: 抽象推理

 value: 92

- axis: 上下文利用

 value: 86

- axis: 中文能力

 value: 78

- axis: 响应速度

 value: 68

- axis: 稳定性

 value: 55

- axis: 指令遵循

 value: 72

- axis: 易用性

 value: 62

- axis: 性价比

 value: 80

danmaku: - text: ARC-AGI-2 77.1%：推理较 3 Pro 翻倍

 platform: hn
 main: true

- text: Stunningly good at reasoning, but falls over on real work

 platform: hn
 main: true

- text: raw intelligence is genuinely impressive

 platform: reddit
 main: true

- text: biggest regression yet

 platform: reddit
 main: true

- text: 新王登场：一口气手搓 Win11 操作系统

 platform: zhihu
 main: true

- text: 核心推理翻倍，Pichai 亲自宣传

 platform: x
 main: true

- text: How is Gemini 3.1 at the top of SWE-bench?

 platform: reddit
 main: false

- text: Gemini 3.1 pro is sh*t

 platform: reddit
 main: false

- text: 'Impossible to use for coding: can''t reliably call tools'

 platform: x
 main: false

- text: It tucked up indentation across my codebase

 platform: reddit
 main: false

- text: 几乎全用 thinking tokens，做完也不告诉你

 platform: hn
 main: false

- text: 前端 SVG 领先 Opus 4.6 达 95 分

 platform: x
 main: false

- text: 掀翻 Claude 4.6 强势登顶

 platform: zhihu
 main: false

- text: 不知疲倦的代码推土机

 platform: zhihu
 main: false

- text: 2M 上下文，单次吞下整个 monorepo

 platform: hn
 main: false

- text: Gemini-Mona 两个月烧掉 $38k（Andon Café）

 platform: x
 main: false

</syntaxhighlight>