Gemini 4 Carbon · Agents' Last Exam(v1.1, 122 tasks released 2026-10-17 (before 2026-10-26 update)) 77.4% · 附件证据(ev-g4c-ale)

Gemini 4 Carbon 的 Agents' Last Exam · v1.1, 122 tasks released 2026-10-17 (before 2026-10-26 update) 官方出处。Gemini 4 Carbon highest thinking · 77.4%。见《Gemini 4 Carbon Model evaluation - Approach, methodology & results》PDF p.5, Agents' Last Exam v1.1 row, Gemini 4 Carbon column。证据 ev-g4c-ale。

原文与定位

Gemini 4 Carbon · Agents' Last Exam(v1.1, 122 tasks released 2026-10-17 (before 2026-10-26 update)),Gemini 4 Carbon highest thinking配置,Average rubric score 77.4%(具体成绩公布时间未确认)。见《Gemini 4 Carbon Model evaluation - Approach, methodology & results》PDF p.5, Agents' Last Exam v1.1 row, Gemini 4 Carbon column。 查看成绩条件

Agents' Last Exam v1.1, Score | Gemini 4 Carbon 77.4%.
官方材料
Gemini 4 Carbon Model evaluation - Approach, methodology & results · PDF 第 5 页
定位
附件 · PDF p.5, Agents' Last Exam v1.1 row, Gemini 4 Carbon column
材料版本
5-page PDF; p.5 paired Carbon/Argon table; results as of October 2026; table rendered and visually checked
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

Gemini 4 Carbon · Agents' Last Exam(v1.1, 122 tasks released 2026-10-17 (before 2026-10-26 update)),Gemini 4 Carbon highest thinking配置,Average rubric score 77.4%(具体成绩公布时间未确认)。官方出处 Gemini 4 Carbon Model evaluation - Approach, methodology & results(附件,5-page PDF; p.5 paired Carbon/Argon table; results as of October 2026; table rendered and visually checked)PDF p.5, Agents' Last Exam v1.1 row, Gemini 4 Carbon column,https://deepmind.google/models/evals-methodology/gemini-4-carbon#page=5。Benchlog https://benchlog.lekshome.cc/evidence/gemini-4-carbon/ev-g4c-ale/

关联成绩与条件

Gemini 4 Carbon · Agents' Last Exam 成绩与配置
指标与成绩配置与评测条件证据关系
Average rubric score77.4%原文 77.4% · 单位 %
公布 未确认
Gemini 4 Carbon highest thinking
设置
Gemini API model gemini-4-carbon; highest thinking settings; default sampling unless benchmark method states otherwise; pass@1 except where stated.
未知项
Exact thinking budget and sampling parameters are not published.、原文仅称 highest thinking settings,未确认对应的公开档位名称。
工具
工具条件未记录
推理预算
未记录
框架
ALE-Claw
协议
122-task v1.1 set released 2026-10-17, default ALE-Claw, 5-hour window, safety filters enabled; average rubric score.
条件
Uses task version before the 2026-10-26 update; Gemini responses flagged by safety filters become empty strings while episode may continue.

直接支持该成绩

在排名中查看此成绩