编排:Glm-5

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: glm-5 tier: T2 stars: 4 composite: 75.8 stats:

 code: 68
 reasoning: 70
 context: 66
 speed: 70
 multimodal: 56
 value: 86

sentiment:

 positive: 55
 mixed: 25
 negative: 20

radar: - axis: 长程任务

 value: 90

- axis: 编程工程

 value: 92

- axis: 抽象推理

 value: 78

- axis: 上下文利用

 value: 82

- axis: 中文能力

 value: 90

- axis: 响应速度

 value: 72

- axis: 稳定性

 value: 76

- axis: 指令遵循

 value: 80

- axis: 易用性

 value: 84

- axis: 性价比

 value: 86

danmaku: - text: GLM 5 is definitely the GOAT

 platform: reddit
 main: true

- text: 测完沉默了，国产什么时候这么能打了

 platform: zhihu
 main: true

- text: Frontier Intelligence Belongs to Everyone

 platform: hn
 main: true

- text: 涨价了，翻了接近一倍

 platform: v2ex
 main: true

- text: 25 分钟一镜到底搓出完整系统

 platform: zhihu
 main: true

- text: 思考模式把正文写进思维链，体感差距很大

 platform: linuxdo
 main: true

- text: I just realised how good GLM 5 is

 platform: reddit
 main: false

- text: GLM 5 is a local GOAT

 platform: reddit
 main: false

- text: Several one shot prompt made mind blowing results!!

 platform: reddit
 main: false

- text: 工程能力站到了 Opus 同一梯队

 platform: zhihu
 main: false

- text: 第一次对国产编程模型说出能打两个字

 platform: zhihu
 main: false

- text: 开源界的 Opus 平替？

 platform: bilibili
 main: false

- text: 等 5.1/5.2 迭代再正式用

 platform: zhihu
 main: false

- text: writes correct diagnoses but ignores its own analysis

 platform: reddit
 main: false

- text: PRO 430，MAX 862，翻了接近一倍

 platform: v2ex
 main: false

- text: GLM 5.2 is Fully Open

 platform: hn
 main: false

</syntaxhighlight>