| 规格 | 数值 |
|---|---|
| 发布方 | Google DeepMind |
| 上下文窗口 | 1M tokens |
来自 Google 发布文章及其基准测试表图片(Gemini 分数为 pass@1;其他模型数据为各厂商自报,评测框架不同)。文章称 2M tokens 即将推出,未给出价格或最大输出。 · 来源
Estimate from the version's published specs, benchmarks and reviews, 0-5; scores are only comparable within one criteria year. Scored as launched (Gemini 2.5 Pro Experimental, 25 Mar 2025). Reasoning null: the Artificial Analysis index current at launch is on a different scale from the 2026 anchors (AA's page now shows 15 on its rescaled index, 16 for the GA model, which would score about 0.5; not used because it is not the launch version); Google's headlines were Humanity's Last Exam 18.8% and #1 on LMArena. Coding null: no Terminal-Bench; Google published SWE-bench Verified 63.8% (custom agent). Context 1M tokens (2M promised) -> 4. Price 4: free experimental at launch; paid preview list price published 4 Apr 2025 (Wayback pricing page 8 Apr 2025): $1.25 in / $10 out per 1M up to 200k prompts, about $3.44 blended (above 200k $2.50/$15). Speed null (none at launch). Modalities 5: text, audio, images, video, code repositories. Reviews: Simon Willison hands-on, Zvi Mowshowitz analysis of user feedback, Analytics Vidhya hands-on (page last updated Aug 2025; original date not shown).[1]
I’m seeing results for Python that feel comparable to my previous favourite Claude 3.7 Sonnet, and appear to be benefitting from Gemini 2.5 Pro’s default reasoning mode and long context.
Gemini 2.5 Pro Experimental is America’s next top large language model. That doesn’t mean it is the best model for everything.
That was simply amazing! This 11 minute long audio clip was processed and summarized in just 18 seconds!
| Putting Gemini 2.5 Pro through its paces · 2025年3月25日 | 0.70 |
|---|---|
| Gemini 2.5 Pro Exp: How to Access, Features, Applications & More · 2025年3月25日 | 0.90 |
| Gemini 2.5 is the New SoTA · 2025年3月28日 | 0.40 |
读者上传的照片,与图钉带来源的媒体分开显示。你可以为吃到或看到的内容评分,在餐厅图钉上还能直接在照片上标记菜品。
第一项始终是图钉的来源。总体可信度是各份资料对上方所用开始和结束时间支持程度的加权平均;资料每比最新的一份旧 180 天,权重减半。
谷歌文章《Gemini 2.5: Our most intelligent AI model》日期为2025年3月25日,并称2.5 Pro Experimental当天即可在Google AI Studio和Gemini应用中使用。
The Wikipedia article lists Gemini 2.5 Pro as released on 25 March 2025, the start of the 2.5 family.
有遗漏或错误吗?用你自己的话说明:能佐证此图钉的链接、不同的开始或结束日期及理由,或缺失、有误的信息。AI 会对照此图钉的来源进行核实,搜索更好的来源,并添加任何支持你说法的页面。图钉自身的来源仍然最重要。AI 也会查看图片:显示的是别的东西或显示效果差的图片会被移到后面或替换。
原标题: Gemini 2.5 Pro Released