GLM-4.6 · CC-Bench(V1.1; 74 real-world coding tests in Claude Code; exact task release unspecified) 56.8% · 附件证据(ev-glm-4-6-completion-cc-bench-v1-1-vs-kimi-k2-0905-win)

GLM-4.6 的 CC-Bench · V1.1; 74 real-world coding tests in Claude Code; exact task release unspecified 官方出处。GLM-4.6 official chart configuration · 56.8%。见《Official GLM-4.6 CC-Bench V1.1 comparison chart》Official CC-Bench V1.1 image, GLM-4.6 vs Kimi-K2-0905 row, Win column。证据 ev-glm-4-6-completion-cc-bench-v1-1-vs-kimi-k2-0905-win。

原文与定位

GLM-4.6 · CC-Bench(V1.1; 74 real-world coding tests in Claude Code; exact task release unspecified),GLM-4.6 official chart configuration配置,Win rate 56.8%(成绩公布 2025-08-11)。见《Official GLM-4.6 CC-Bench V1.1 comparison chart》Official CC-Bench V1.1 image, GLM-4.6 vs Kimi-K2-0905 row, Win column。 查看成绩条件

GLM-4.6 vs Kimi-K2-0905: 56.8% Win, 28.3% Tie, 14.9% Loss.
官方材料
Official GLM-4.6 CC-Bench V1.1 comparison chart
定位
附件 · Official CC-Bench V1.1 image, GLM-4.6 vs Kimi-K2-0905 row, Win column
材料版本
Direct Z.AI documentation image; all four win/tie/loss rows visually read; local file 本地留存副本.
版本公布
未确认
披露时间
未确认 · 首发/后补未确认
采集时间
历史依据
暂无可靠历史版本依据
引用文本

GLM-4.6 · CC-Bench(V1.1; 74 real-world coding tests in Claude Code; exact task release unspecified),GLM-4.6 official chart configuration配置,Win rate 56.8%(成绩公布 2025-08-11)。官方出处 Official GLM-4.6 CC-Bench V1.1 comparison chart(附件,Direct Z.AI documentation image; all four win/tie/loss rows visually read; local file 本地留存副本.)Official CC-Bench V1.1 image, GLM-4.6 vs Kimi-K2-0905 row, Win column,https://cdn.bigmodel.cn/markdown/1759212585375glm-4.6-2.jpeg?attname=glm-4.6-2.jpeg。Benchlog https://benchlog.lekshome.cc/evidence/glm-4-6/ev-glm-4-6-completion-cc-bench-v1-1-vs-kimi-k2-0905-win/

关联成绩与条件

GLM-4.6 · CC-Bench 成绩与配置
指标与成绩配置与评测条件证据关系
Win rate56.8%原文 56.8% · 单位 %
公布
GLM-4.6 official chart configuration
设置
GLM-4.6; chart column is not labelled w/ Tools. Model serving, effort and precision are not disclosed.
未知项
Effort and serving endpoint are unknown.、Chart values have no printed units.
工具
工具条件未记录
推理预算
未记录
框架
Claude Code
协议
CC-Bench V1.1; official docs describe 74 real-world coding tests in the Claude Code environment; comparison against Kimi-K2-0905.
条件
GLM-4.6; task-level harness details and serving precision are not stated in the chart.

直接支持该成绩

在排名中查看此成绩