Gemini 2.5 Pro (GA) · VideoMME(Full test set; audio + visual + subtitles) 86.9 · 附件证据(t6-videomme-subtitles)

Gemini 2.5 Pro (GA) 的 VideoMME · Full test set; audio + visual + subtitles 官方出处。Gemini 2.5 Pro stable API · 86.9。见《Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities》PDF p.15, Table 6, VideoMME audio+visual+subtitles row; Table 11 says full test set。证据 t6-videomme-subtitles。

原文与定位

Gemini 2.5 Pro (GA) · VideoMME(Full test set; audio + visual + subtitles),Gemini 2.5 Pro stable API配置,Accuracy 86.9(具体成绩公布时间未确认)。见《Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities》PDF p.15, Table 6, VideoMME audio+visual+subtitles row; Table 11 says full test set。 查看成绩条件

VideoMME 77.3 79.8 72.5 78.8 81.5 86.9 79.6
官方材料
Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities · PDF 第 15 页
定位
附件 · PDF p.15, Table 6, VideoMME audio+visual+subtitles row; Table 11 says full test set
材料版本
73-page PDF; embedded results include Gemini 2.5 Pro GA up to version 06-17; PDF metadata creation 2025-10-16 is not treated as publication date
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 2.5 Pro (GA) · VideoMME(Full test set; audio + visual + subtitles),Gemini 2.5 Pro stable API配置,Accuracy 86.9(具体成绩公布时间未确认)。官方出处 Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities(附件,73-page PDF; embedded results include Gemini 2.5 Pro GA up to version 06-17; PDF metadata creation 2025-10-16 is not treated as publication date)PDF p.15, Table 6, VideoMME audio+visual+subtitles row; Table 11 says full test set,https://storage.googleapis.com/deepmind-media/gemini/gemini_v2_5_report.pdf#page=15。Benchlog https://benchlog.lekshome.cc/evidence/gemini-2-5-pro/t6-videomme-subtitles/

关联成绩与条件

Gemini 2.5 Pro (GA) · VideoMME 成绩与配置
指标与成绩配置与评测条件证据关系
Accuracy86.9原文 86.9 · 单位 未标注
公布 未确认
Gemini 2.5 Pro stable API
设置
AI Studio API model id gemini-2.5-pro; default sampling settings. Technical report states Gemini scores are pass@1 and single-attempt unless specified. Aider Polyglot averages 3 trials; SWE-bench uses the internal agentic harness; Vibe-Eval uses Gemini as judge.
未知项
Thinking budget and any non-default settings for individual evaluations are not stated except where noted in each protocol.
工具
工具条件未记录
推理预算
未记录
框架
未记录
协议
0-shot; audio+visual+subtitles; full test set; max 1024 frames.
条件
Video sampled at 1 fps; modality and subset are stated in the report.

直接支持该成绩

在排名中查看此成绩