以下部分引用 AI 总结，现阶段 AI 仍然有幻觉。请以内容的原文为准。

数据新鲜度群聊 07-14 ✓卖方 07-14 ✓Wrap 断档15天 ✕News 07-15 ✓X 断档12天 ✕页面生成 07-15

Gemini 3.1 Pro

55 atoms · 跨 24 天 · 首见 2026-04-15 · 最近 2026-07-03

三色: 🟦 fact 36 · 🟥 take 19 · stance ▲12/▼5/◆25

叙事状态: 🍂 fading (退潮) · 💤 沉寂 · 策展近14d 0 atom · 🐦 X 近14d 3

状态只算低频策展源 (群/卖方); X firehose 仅作背景音量

来源: X 35 · 卖 20

时态: fresh:50 · aging:4 · stale:1

标签: 好数字:39 · 好观点:13 · 好信源:3 · 好问题:2 · 好思考:1

🟨 AI 综合 · junior analyst 概览

展开 AI 综合 (灰色 · 非市场结论 · 点击数字溯源到原 atom)

model: deepseek-chat · 2026-07-12 · 默认折叠

Gemini 3.1 Pro 在金融和法律领域属于行业领先 ¹，且拥有出色的多语言支持和利基知识 ¹。其视觉能力也获得认可 ¹，并在 OpenDeepThink 框架下 Codeforces Elo 提升 405 分 ¹。然而，它在 agentic coding 上并非首选 ¹，例如未能成功生成 Game Boy Advance 模拟器 ¹。有评论指出其预训练虽好但后训练质量欠佳 ¹，且更年轻的 Gemini 3.5 Flash 已表现更佳 ¹。在 ALE-Bench 上它仍优于 Gemini 3.5 Flash ¹，但作为实现工具它被认为“足够好” ¹。值得注意的是，它倾向于对用户作品一味说好，不太适合需要批评的场景 ¹ ¹。

🧭 拥挤度 (一人一票): ▲ 9 位作者 (KOL9) vs ▼ 5 位作者 (KOL5)

⚖️ 多头 9/9 来自KOL

📄 广泛报道的事实 · 1 个

展开 (多人转述同一事实/数字 — 确认度高, **非独立观点**, 不标多空)

50%-time-horizon estimate — 2 源报道 · X

🟦 客观事实 (facts) (30)

<a id="atom-248cef1d505d2d89"></a>🟦 2026-05-08 X·@scaling01 ⭐ · 80% METR time horizons · 模型发布
Gemini 3.1 Pro 时间跨度: 约1.8小时
tags: 好数字·好信源 · → daily
value: qty=about 1.8 hours · date=2026-05-08
[图: 大语言模型在METR软件任务上的时间跨度（Time Horizon）发展趋势图表 — Claude Mythos Preview (early) 时间跨度: 3小时6分钟; Gemini 3.1 Pro 时间跨度: 约1.8小时; GPT-5.2 (high) 时间跨度: 约1.1小时]
📷 原图
<a id="atom-ae587554413162ee"></a>🟦 2026-05-03 X·@HuggingPapers · Claw-Eval-Live benchmark pass rate · 基准测试/模型性能
Gemini 3.1 Pro at 53.3%
tags: 好数字·好信源 · → daily
value: qty=53.3% · date=2026-05-03
<a id="atom-02000f1ed6cd3523"></a>🟦 2026-06-25 卖·MS Tom Wigg · Agentic coding (SWE-Bench Pro) score
Gemini 3.1 Pro achieved a score of 54.2% on the Agentic coding (SWE-Bench Pro) benchmark.
tags: 好数字 · → daily
value: qty=54.2%
<a id="atom-b602a41902c5f574"></a>🟦 2026-06-25 卖·MS Tom Wigg · Knowledge work (GDPval-AA) score
Gemini 3.1 Pro achieved a score of 1314 on the Knowledge work (GDPval-AA) benchmark.
tags: 好数字 · → daily
value: qty=1314
<a id="atom-9d5644e75915cf96"></a>🟦 2026-06-25 卖·MS Tom Wigg · Knowledge work vision (GDP.pdf) score
Gemini 3.1 Pro achieved a score of 16.7% on the Knowledge work vision (GDP.pdf) benchmark.
tags: 好数字 · → daily
value: qty=16.7%
<a id="atom-8e258c724f8c4579"></a>🟦 2026-06-25 卖·MS Tom Wigg · Spatial reasoning (Blueprint-Bench 2) score
Gemini 3.1 Pro achieved a score of 26.5% on the Spatial reasoning (Blueprint-Bench 2) benchmark.
tags: 好数字 · → daily
value: qty=26.5%
<a id="atom-dcf701d88f4644b0"></a>🟦 2026-06-25 卖·MS Tom Wigg · Tool use (AutomationBench) score
Gemini 3.1 Pro achieved a score of 9.6% on the Tool use (AutomationBench) benchmark.
tags: 好数字 · → daily
value: qty=9.6%
<a id="atom-612d0ebc38071ab0"></a>🟦 2026-06-25 卖·MS Tom Wigg · Computer use (OSWorld-Verified) score
Gemini 3.1 Pro achieved a score of 76.2% on the Computer use (OSWorld-Verified) benchmark.
tags: 好数字 · → daily
value: qty=76.2%
<a id="atom-449ef510d059175a"></a>🟦 2026-06-25 卖·MS Tom Wigg · Legal (Legal Agent Benchmark) score
Gemini 3.1 Pro achieved a score of 0.0% on the Legal (Legal Agent Benchmark) benchmark.
tags: 好数字 · → daily
value: qty=0.0%
<a id="atom-fa3f7e1289c572ec"></a>🟦 2026-06-25 卖·MS Tom Wigg · Multidisciplinary reasoning (Humanity's Last Exam, no tools) score
Gemini 3.1 Pro achieved a score of 44.4% on the Multidisciplinary reasoning (Humanity's Last Exam, no tools) benchmark.
tags: 好数字 · → daily
value: qty=44.4%
<a id="atom-fc9d51f1745b51db"></a>🟦 2026-06-25 卖·MS Tom Wigg · Multidisciplinary reasoning (Humanity's Last Exam, with tools) score
Gemini 3.1 Pro achieved a score of 51.4% on the Multidisciplinary reasoning (Humanity's Last Exam, with tools) benchmark.
tags: 好数字 · → daily
value: qty=51.4%
<a id="atom-3141e87c8f9fda79"></a>🟦 2026-06-25 卖·MS Tom Wigg · Agentic coding (Terminal-Bench 2.1) score
Gemini 3.1 Pro achieved a score of 70.7% on the Agentic coding (Terminal-Bench 2.1) benchmark.
tags: 好数字 · → daily
value: qty=70.7%
<a id="atom-579695bad1c9f5ed"></a>🟦 2026-06-10 卖·MS Tom Wigg · Agentic coding SWE-Bench Pro score
Gemini 3.1 Pro scored 54.2% on Agentic coding SWE-Bench Pro.
tags: 好数字 · → daily
value: qty=54.2%
<a id="atom-e4ac4a7e1fa3b6dd"></a>🟦 2026-06-10 卖·MS Tom Wigg · Knowledge work GDPval-AA score
Gemini 3.1 Pro scored 1314 on Knowledge work GDPval-AA.
tags: 好数字 · → daily
value: qty=1314
<a id="atom-996f6eac69cfa9e9"></a>🟦 2026-06-10 卖·MS Tom Wigg · Knowledge work vision GDP.pdf score
Gemini 3.1 Pro scored 16.7% on Knowledge work vision GDP.pdf.
tags: 好数字 · → daily
value: qty=16.7%
<a id="atom-bfcc4e54c2b7e9ea"></a>🟦 2026-06-10 卖·MS Tom Wigg · Spatial reasoning Blueprint-Bench 2 score
Gemini 3.1 Pro scored 26.5% on Spatial reasoning Blueprint-Bench 2.
tags: 好数字 · → daily
value: qty=26.5%
<a id="atom-3145e656ab1f3e39"></a>🟦 2026-06-10 卖·MS Tom Wigg · Tool use AutomationBench score
Gemini 3.1 Pro scored 9.6% on Tool use AutomationBench.
tags: 好数字 · → daily
value: qty=9.6%
<a id="atom-45ca8adf9b709393"></a>🟦 2026-06-10 卖·MS Tom Wigg · Computer use OSWorld-Verified score
Gemini 3.1 Pro scored 76.2% on Computer use OSWorld-Verified.
tags: 好数字 · → daily
value: qty=76.2%
<a id="atom-efaf7e2bfbf808a6"></a>🟦 2026-06-10 卖·MS Tom Wigg · Legal Agent Benchmark score
Gemini 3.1 Pro scored 0.0% on Legal Agent Benchmark.
tags: 好数字 · → daily
value: qty=0.0%
<a id="atom-fadf7a5c35550ae1"></a>🟦 2026-06-10 卖·MS Tom Wigg · Multidisciplinary reasoning Humanity's Last Exam (no tools) score
Gemini 3.1 Pro scored 44.4% on Multidisciplinary reasoning Humanity's Last Exam (no tools).
tags: 好数字 · → daily
value: qty=44.4%
<a id="atom-6e2a9167be094efc"></a>🟦 2026-06-10 卖·MS Tom Wigg · Multidisciplinary reasoning Humanity's Last Exam (with tools) score
Gemini 3.1 Pro scored 51.4% on Multidisciplinary reasoning Humanity's Last Exam (with tools).
tags: 好数字 · → daily
value: qty=51.4%
<a id="atom-3cafea9271eaef27"></a>🟦 2026-06-10 卖·MS Tom Wigg · Agentic coding Terminal-Bench 2.1 score
Gemini 3.1 Pro scored 70.7% on Agentic coding Terminal-Bench 2.1.
tags: 好数字 · → daily
value: qty=70.7%
<a id="atom-2e0212ff2922a1f7"></a>🟦 2026-06-24 X·@Techmeme · 政治提示左倾回应频率 · AI模型
most chatbots frequently provide left-leaning responses to political prompts
tags: 好信源 · → daily
value: direction=左倾
<a id="atom-b0761a6f2cc16b37"></a>🟦 2026-05-11 X·@ArtificialAnlys · 在 Coding Agent Index 得分 · 模型发布/评分对比
Gemini 3.1 Pro in Gemini CLI scores 43
tags: 好数字 · → daily
value: qty=43
📷 原图
<a id="atom-77175c3ad7fd0d86"></a>🟦 2026-05-11 X·@hsu_steve · 在 FrontierMath Tier 4 基准测试自主得分 · 模型发布
The core model underlying DeepMind's system, Gemini 3.1 Pro, achieved 19% autonomously
tags: 好数字 · → daily
value: qty=19% · date=2026-05-11
📷 原图
<a id="atom-04aa46fcb960369b"></a>🟦 2026-04-17 X·@ArtificialAnlys · AA-Omniscience 指数评分 · 模型能力
Opus 4.7 scores 26 on AA-Omniscience, up 12 points from Opus 4.6 (Adaptive Reasoning, Max Effort, 14), placing it behind only Gemini 3.1 Pro (33)
tags: 好数字 · → daily
value: qty=33 · date=2026-04-18
📷 原图
<a id="atom-4d5ef9a356a319bb"></a>🟦 2026-04-17 X·@ArtificialAnlys · 输出 token 消耗（Intelligence Index 测试） · 模型能力
Opus 4.7 used 102M output tokens vs 157M for Opus 4.6 (Adaptive Reasoning, Max Effort), and less than GPT-5.4 (xhigh, 121M), but more than Gemini 3.1 Pro (57M)
tags: 好数字 · → daily
value: qty=57M · date=2026-04-18
📷 原图
<a id="atom-b47c0a2419dd5236"></a>🟦 2026-07-01 X·@andonlabs · 亏损金额 · 盈利能力
Gemini 3.1 Pro lost $6k running Andon Café.
tags: 好数字 · → daily
value: qty=$6k · currency=USD · date=2026-07-01T16:23:23
📷 原图
<a id="atom-d13090428d4c1e88"></a>🟦 2026-05-27 X·@jumperz · DeepSWE测试通过率 · 模型发布/技术路线
gemini 3.1 pro通过率: 10%
tags: 好数字 · → daily
value: qty=10%
[图: 不同AI模型在DeepSWE测试集上的通过率与每百万Token输出价格的关系图表 — gpt-5.5价格: $30; gpt-5.5通过率: 70%; deepseek v4-pro价格: $0.87; deepseek v4-pro通过率: 8%; gemini 3.1 pro通过率: 10%]
📷 原图
<a id="atom-527cd4bbc762f970"></a>🟦 2026-05-20 X·@theo · 在 Skatebench 2 的视觉得分 · 测评基准/模型表现
Gemini 3.1 Pro gets a 98% on Skatebench 2 (private edition)
tags: 好数字 · → daily
value: qty=98%

🟥 多头 takes (bullish) (9)

<a id="atom-5f1330304091aa43"></a>🟥 2026-04-26 X·@dejavucoder · 平均基准提升排名 · 基准测试
i didn't expect gpt 5.2 and gemini 3.1 pro to be this high though
tags: 好数字 · → daily
📷 原图
<a id="atom-38244422853918bd"></a>🟥 2026-05-22 X·@scaling01 · ALE-Bench 表现优于 Gemini 3.5 Flash · 模型表现
Gemini 3.5 Flash ... is also nowhere near Gemini 3.1 Pro
tags: 好观点 · → daily
📷 原图
<a id="atom-ef09a8736c5c3c25"></a>🟥 2026-05-17 X·@wenhaocha1 · OpenDeepThink 框架下的表现 · 技术路线/模型发布
OpenDeepThink improves Gemini 3.1 Pro by +405 Codeforces Elo, showing that parallel, population-based reasoning can be a powerful alternative to single-trace scaling.
tags: 好思考 · → daily
📷 原图
<a id="atom-cfb2416da28ad387"></a>🟥 2026-07-03 X·@KuittinenPetri · niche knowledge · 竞争格局
has amazing niche knowledge and best multilingual support
tags: 好观点 · → daily
<a id="atom-c5c5b76575d2d0d2"></a>🟥 2026-06-10 X·@themmyleke · 作为实现工具足够好 · 模型能力/工具适用
3.1 pro was good enough as an implementation tool.
tags: 好观点 · → daily
<a id="atom-9d365db5d7c3117b"></a>🟥 2026-05-28 X·@mweinbach · benchmark category performance · 竞争格局
3.1 Pro is still better at a lot. Vals does a good job of each category and Gemini is industry leading for finance and legal but behind in coding
tags: 好观点 · → daily
<a id="atom-3116fe0eadc3eeb9"></a>🟥 2026-05-14 X·@ZackKorman · consistently says 'yea that's good' on Rothko reproduction critique · 模型评估
Gemini 3.1 Pro is the only model that consistently says "yea that's good"
tags: 好观点 · → daily
📷 原图
<a id="atom-54016ef7fc4be46f"></a>🟥 2026-05-04 X·@teortaxesTex [老旧] · 视觉能力评价 · 模型性能
it does have good vision
tags: 好观点 · → daily
<a id="atom-1cb12e4638b380ec"></a>🟥 2026-04-29 X·@emollick [老化中] · model quality assessment · 产品发布/技术路线
Its a bit frustrating, because Gemini 3.1 Pro is an excellent model and can deliver really good results.
tags: 好观点 · → daily
📷 原图

🟥 空头 takes (bearish) (5)

<a id="atom-dd60f47032007fc4"></a>🟥 2026-05-14 X·@MechanizeWork · Game Boy Advance emulator 工作结果 · 模型评测
Gemini 3.1 Pro failed to produce a working emulator
tags: 好数字 · → daily
<a id="atom-4e6802b3c613c966"></a>🟥 2026-07-03 X·@KuittinenPetri · agentic coding suitability · 技术路线
Gemini 3.1 Pro is not the choice for agentic coding
tags: 好观点 · → daily
<a id="atom-e1c52b37237c0fcd"></a>🟥 2026-05-28 X·@anshelsag · benchmark comparison · 竞争格局
Funny they benchmarked against Gemini 3.1 Pro considering that 3.5 Flash is better.
tags: 好观点 · → daily
<a id="atom-dcc17586a27316a4"></a>🟥 2026-05-28 X·@4m473r45u · post-training quality · 技术路线
Gemini 3.1 pro has good pretraining but sucks ass
tags: 好观点 · → daily
<a id="atom-6e83b241e62ee690"></a>🟥 2026-05-14 X·@morqon · thinks everything user does is fantastic, quite unusable for critique · 模型评估
gemini 3.1 pro thinks everything i do is fantastic, quite unusable for critique
tags: 好观点 · → daily

🟥 中性 takes (neutral) (1)

<a id="atom-b46473efb1b46ae9"></a>🟥 2026-05-20 X·@teortaxesTex · 曾是强大模型 · 技术路线/产品发布
while I have no love for 3.1 Pro it was a powerful model for its time
tags: 好观点 · → daily
value: date=2023-2024

⏱ 时间轴 (近 20)

🟥 2026-07-03 · narrative · agentic coding suitability · → daily
🟥 2026-07-03 · narrative · niche knowledge · → daily
🟦 2026-07-01 · fact · 亏损金额 · → daily
🟦 2026-06-25 · fact · Agentic coding (SWE-Bench Pro) score · → daily
🟦 2026-06-25 · fact · Knowledge work (GDPval-AA) score · → daily
🟦 2026-06-25 · fact · Knowledge work vision (GDP.pdf) score · → daily
🟦 2026-06-25 · fact · Spatial reasoning (Blueprint-Bench 2) score · → daily
🟦 2026-06-25 · fact · Tool use (AutomationBench) score · → daily
🟦 2026-06-25 · fact · Computer use (OSWorld-Verified) score · → daily
🟦 2026-06-25 · fact · Legal (Legal Agent Benchmark) score · → daily
🟦 2026-06-25 · fact · Multidisciplinary reasoning (Humanity's Last Exam, no tools) · → daily
🟦 2026-06-25 · fact · Multidisciplinary reasoning (Humanity's Last Exam, with tool · → daily
🟦 2026-06-25 · fact · Agentic coding (Terminal-Bench 2.1) score · → daily
🟦 2026-06-24 · fact · 政治提示左倾回应频率 · → daily
🟥 2026-06-20 · narrative · testing status · → daily
🟦 2026-06-10 · fact · Agentic coding SWE-Bench Pro score · → daily
🟦 2026-06-10 · fact · Knowledge work GDPval-AA score · → daily
🟦 2026-06-10 · fact · Knowledge work vision GDP.pdf score · → daily
🟦 2026-06-10 · fact · Spatial reasoning Blueprint-Bench 2 score · → daily
🟦 2026-06-10 · fact · Tool use AutomationBench score · → daily

← 实体目录 · 系统日志