GLM-4.5 · CC-Bench(52 programming/development tasks across six domains; task release unspecified) 90.6% · 附件证据(ev-glm-4-5-completion-cc-bench-tool-call-success)

GLM-4.5 的 CC-Bench · 52 programming/development tasks across six domains; task release unspecified 官方出处。GLM-4.5 reported · 90.6%。见《GLM-4.5 Technical Report (PDF)》Technical report, Section 4.3.2, Figure 13, PDF page 19; average tool-calling success rate。证据 ev-glm-4-5-completion-cc-bench-tool-call-success。

原文与定位

GLM-4.5 · CC-Bench(52 programming/development tasks across six domains; task release unspecified),GLM-4.5 reported配置,Tool-calling success rate 90.6%(成绩公布 2025-08-11)。见《GLM-4.5 Technical Report (PDF)》Technical report, Section 4.3.2, Figure 13, PDF page 19; average tool-calling success rate。 查看成绩条件

Average Tool Calling Success Rate Comparison / Success Rate(%) / GLM-4.5 / 90.6%; Figure 13: Average tool calling success rate and token usage per interaction across different models on CC-Bench.
官方材料
GLM-4.5 Technical Report (PDF) · PDF 第 19 页
定位
附件 · Technical report, Section 4.3.2, Figure 13, PDF page 19; average tool-calling success rate
材料版本
Official 26-page PDF; full report read 26/26 pages, including Tables 3–6, methods and figure/table pages; collected 2026-11-17.
版本公布
未确认
披露时间
未确认 · 后补披露
采集时间
历史依据
暂无可靠历史版本依据
引用文本

GLM-4.5 · CC-Bench(52 programming/development tasks across six domains; task release unspecified),GLM-4.5 reported配置,Tool-calling success rate 90.6%(成绩公布 2025-08-11)。官方出处 GLM-4.5 Technical Report (PDF)(附件,Official 26-page PDF; full report read 26/26 pages, including Tables 3–6, methods and figure/table pages; collected 2026-11-17.)Technical report, Section 4.3.2, Figure 13, PDF page 19; average tool-calling success rate,https://arxiv.org/pdf/2508.06471#page=19。Benchlog https://benchlog.lekshome.cc/evidence/glm-4-5/ev-glm-4-5-completion-cc-bench-tool-call-success/

关联成绩与条件

GLM-4.5 · CC-Bench 成绩与配置
指标与成绩配置与评测条件证据关系
Tool-calling success rate90.6%原文 90.6% · 单位 %
公布
GLM-4.5 reported
设置
GLM-4.5; 355B total, 32B active parameters.
未知项
API/hosted endpoint versus local inference and exact weight precision are not specified in the report.
工具
工具条件未记录
推理预算
未记录
框架
未记录
协议
Average tool-calling success rate on CC-Bench; 52 programming tasks across six domains.
条件
GLM-4.5; figure compares with Claude Sonnet 4, Kimi K2 and Qwen3-Coder.

直接支持该成绩

在排名中查看此成绩