GPT-5.4 nano · MMMU-Pro(Standard/Vision 等格式未在所选材料中注明) 69.5% · 主文证据(ev-gpt54-nano-mmmu-python)

GPT-5.4 nano 的 MMMU-Pro · Standard/Vision 等格式未在所选材料中注明 官方出处。GPT-5.4 nano (xhigh) · 69.5%。见《Introducing GPT-5.4 mini and nano | OpenAI》Table 5 / row “MMMUPro w/ Python” / column “GPT-5.4 nano (xhigh)”。证据 ev-gpt54-nano-mmmu-python。

原文与定位

GPT-5.4 nano · MMMU-Pro(Standard/Vision 等格式未在所选材料中注明),GPT-5.4 nano (xhigh)配置,Score 69.5%(成绩公布 2026-03-17)。见《Introducing GPT-5.4 mini and nano | OpenAI》Table 5 / row “MMMUPro w/ Python” / column “GPT-5.4 nano (xhigh)”。 查看成绩条件

MMMUPro w/ Python | GPT-5.4 nano (xhigh): 69.5%
官方材料
Introducing GPT-5.4 mini and nano | OpenAI
定位
主文 · Table 5 / row “MMMUPro w/ Python” / column “GPT-5.4 nano (xhigh)”
材料版本
Saved Wayback capture from 2026-03-17T17:52:20Z; the article body and all six result tables were read. Current page revision is unknown.
版本公布
披露时间
· 已有首发依据
采集时间
历史依据
The saved capture is the 2026-03-17 17:52:20Z replay. Later replay attempts on 2026-11-16 returned HTTP 502/429; the saved body was used for the launch-day review.
引用文本

GPT-5.4 nano · MMMU-Pro(Standard/Vision 等格式未在所选材料中注明),GPT-5.4 nano (xhigh)配置,Score 69.5%(成绩公布 2026-03-17)。官方出处 Introducing GPT-5.4 mini and nano | OpenAI(主文,Saved Wayback capture from 2026-03-17T17:52:20Z; the article body and all six result tables were read. Current page revision is unknown.)Table 5 / row “MMMUPro w/ Python” / column “GPT-5.4 nano (xhigh)”,https://openai.com/index/introducing-gpt-5-4-mini-and-nano/。Benchlog https://benchlog.lekshome.cc/evidence/gpt-5-4-nano/ev-gpt54-nano-mmmu-python/

关联成绩与条件

GPT-5.4 nano · MMMU-Pro 成绩与配置
指标与成绩配置与评测条件证据关系
Score69.5%原文 69.5% · 单位 %
公布
GPT-5.4 nano (xhigh)
设置
The official comparison table labels GPT-5.4 nano (xhigh).
未知项
未另记
工具
允许工具
推理预算
xhigh
框架
未记录
协议
MMMU-Pro with Python; exact format and tool harness are not stated.
条件
GPT-5.4 nano (xhigh); evaluation conducted in a research environment.

直接支持该成绩

在排名中查看此成绩