Gemini 3 Pro · MMMU-Pro(Standard and Vision formats) 81% · 主文证据(main-mmmu-pro-launch)

Gemini 3 Pro 的 MMMU-Pro · Standard and Vision formats 官方出处。Gemini 3 Pro API default · 81%。见《A new era of intelligence with Gemini 3 — launch-day article》Release article / multimodal reasoning paragraph。证据 main-mmmu-pro-launch。

原文与定位

Gemini 3 Pro · MMMU-Pro(Standard and Vision formats),Gemini 3 Pro API default配置,Accuracy 81%(具体成绩公布时间未确认)。见《A new era of intelligence with Gemini 3 — launch-day article》Release article / multimodal reasoning paragraph。 查看成绩条件

81% on MMMU-Pro
官方材料
A new era of intelligence with Gemini 3 — launch-day article
定位
主文 · Release article / multimodal reasoning paragraph
材料版本
Official article archived 2025-11-18 16:07:15 UTC; includes independently recovered launch-day benchmark images.
版本公布
披露时间
· 已有首发依据
采集时间
历史依据
实际取得官方主文首发当天历史回放 https://web.archive.org/web/20251118160715id_/https://blog.google/products/gemini/gemini-3/;Memento-Datetime Tue, 18 Nov 2025 16:07:15 GMT。主表与 Deep Think 图同日16:08:44回放、Vending-Bench图同日16:09:03回放均已下载并视觉核查,不依赖现行图像字节。
引用文本

Gemini 3 Pro · MMMU-Pro(Standard and Vision formats),Gemini 3 Pro API default配置,Accuracy 81%(具体成绩公布时间未确认)。官方出处 A new era of intelligence with Gemini 3 — launch-day article(主文,Official article archived 2025-11-18 16:07:15 UTC; includes independently recovered launch-day benchmark images.)Release article / multimodal reasoning paragraph,https://blog.google/products/gemini/gemini-3/。Benchlog https://benchlog.lekshome.cc/evidence/gemini-3-pro/main-mmmu-pro-launch/

关联成绩与条件

Gemini 3 Pro · MMMU-Pro 成绩与配置
指标与成绩配置与评测条件证据关系
Accuracy81%原文 81.0% · 单位 %
公布 未确认
Gemini 3 Pro API default
设置
Gemini API model id gemini-3-pro-preview; default sampling settings. Gemini results are pass@1 and single-attempt unless specified; smaller benchmarks average multiple trials.
未知项
The evaluation table does not specify a thinking budget or inference-time compute setting.
工具
禁止工具
推理预算
未记录
框架
未记录
协议
No tools; average of Standard (10 options) and Vision settings.
条件
Multimodal understanding and reasoning.

直接支持该成绩

在排名中查看此成绩