Google DeepMind 模型卡,截至 2026 年 2 月的结果。 · 来源
单次尝试;该卡没有给出该基准测试的克劳德数据。 · 来源
| 物品 | 价值 |
|---|---|
| 输入 | 文字、图像、音频、视频;上下文窗口最多 1M 个令牌 |
| 输出 | 文本,64K 令牌 |
| 基于 | 双子座 3 专业版 |
| 人类的最后考试(无工具) | 44.4%(双子座 3 Pro 37.5%) |
| GPQA 钻石级(无工具) | 94.3% |
| SWE-Bench 验证(单次尝试) | 80.6% |
| 终端工作台 2.0 (Terminus-2) | 68.5% |
| tau2-bench 电信 | 99.3% |
| 浏览比较 | 85.9% |
| MCP图集 | 69.2% |
| MRCR v2 8 针,1M 点 | 26.3% |
Google 自己的数据,截至 2026 年 2 月的结果。该卡未列出 API 价格。 · 来源
Partial outage now. Google Gemini is currently experiencing a partial outage
Unofficial, from StatusGator. Uptime is the share of the window not marked down (559 min down, 3277 min degraded); it covers the whole service, not this version alone.
Estimate from the version's published specs, benchmarks and reviews, 0-5; scores are only comparable within one criteria year. Scored as launched (19 Feb 2026, preview). Reasoning: Artificial Analysis Intelligence Index 57 at launch (v4.0; four points ahead of Claude Opus 4.6; The Decoder 21 Feb 2026). AA's pages now show lower numbers (30, 46) after later index versions, not used. Coding null: no Terminal-Bench 4.0; Google published Terminal-Bench 2.0 68.5% (Terminus-2) and SWE-Bench Verified 80.6%. Context 1M input, 64k output (DeepMind model card). Price $2 in / $12 out per 1M up to 200k = $4.50 blended (AA, Google). Speed null: no launch-time speed found (AA now lists 132 tokens/s, undated). Input text, image, audio, video.[1]
The model ranks first in six of ten categories, including agent-based coding, knowledge, scientific reasoning, and physics.
While the leap in logical reasoning and coding benchmarks is impressive, the 'soul' of the model — its emotional depth, empathy, creative flexibility, and nuance — seems to have been significantly reduced.
It is not necessarily faster, and in some modes, it is intentionally slower, taking a moment to chew on a problem before spitting out an answer.
Winner: Gemini wins for providing phrasing I'd actually use because it feels the most natural and immediately usable.
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 3.1 Pro: A smarter model for your most complex tasks》日期为2026年2月19日,并称当天起向开发者、企业和消费者推送。
The Wikipedia article lists Gemini 3.1 Pro as released in preview on 19 February 2026.
DeepMind's model card gives the benchmark table against the model's predecessor and named rivals.
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 3.1 Pro Released
| 基准测试 | Gemini 3.1 Pro | Gemini 3 Pro | Sonnet 4.6 | Opus 4.6 | GPT-5.2 | GPT-5.3-Codex |
|---|---|---|---|---|---|---|
| Humanity's Last Exam(学术推理,完整集,文本+多模态,不使用工具) | 44.4% | 37.5% | 33.2% | 40.0% | 34.5% | — |
| Humanity's Last Exam(学术推理,完整集,文本+多模态,搜索(屏蔽名单)+代码) | 51.4% | 45.8% | 49.0% | 53.1% | 45.5% | — |
| ARC-AGI-2(抽象推理谜题,ARC Prize验证) | 77.1% | 31.1% | 58.3% | 68.8% | 52.9% | — |
| GPQA Diamond(科学知识,不使用工具) | 94.3% | 91.9% | 89.9% | 91.3% | 92.4% | — |
| Terminal-Bench 2.0(智能体终端编程,Terminus-2框架) | 68.5% | 56.9% | 59.1% | 65.4% | 54.0% | 64.7% |
| Terminal-Bench 2.0(智能体终端编程,其他自报最佳框架) | — | — | — | — | 62.2% | 77.3% |
| SWE-Bench Verified(智能体编程,单次尝试) | 80.6% | 76.2% | 79.6% | 80.8% | 80.0% | — |
| SWE-Bench Pro (Public)(多样化智能体编程任务,单次尝试) | 54.2% | 43.3% | — | — | 55.6% | 56.8% |
| LiveCodeBench Pro(来自Codeforces、ICPC和IOI的竞赛编程题,Elo) | 2887 | 2439 | — | — | 2393 | — |
| SciCode(科研编程) | 59% | 56% | 47% | 52% | 52% | — |
| APEX-Agents(长周期专业任务) | 33.5% | 18.4% | — | 29.8% | 23.0% | — |
| GDPval-AA Elo(专家任务) | 1317 | 1195 | 1633 | 1606 | 1462 | — |
| τ2-bench(智能体与工具使用,零售) | 90.8% | 85.3% | 91.7% | 91.9% | 82.0% | — |
| τ2-bench(智能体与工具使用,电信) | 99.3% | 98.0% | 97.9% | 99.3% | 98.7% | — |
| MCP Atlas(使用MCP的多步骤工作流) | 69.2% | 54.1% | 61.3% | 59.5% | 60.6% | — |
| BrowseComp(智能体搜索,搜索+Python+浏览) | 85.9% | 59.2% | 74.7% | 84.0% | 65.8% | — |
| MMMU-Pro(多模态理解与推理,不使用工具) | 80.5% | 81.0% | 74.5% | 73.9% | 79.5% | — |
| MMMLU(多语言问答) | 92.6% | 91.8% | 89.3% | 91.1% | 89.6% | — |
| MRCR v2 (8-needle)(长上下文表现,128k(平均)) | 84.9% | 77.0% | 84.9% | 84.0% | 83.8% | — |
| MRCR v2 (8-needle)(长上下文表现,1M(逐点)) | 26.3% | 26.3% | — | — | — | — |