【AI 英文奏折】2026年11月18日|模型迭代:Astra 6.2 日常自动化

每日奏折13小时前发布 tianming
18 0 0

共收录 18 篇深度内容(其中 3 条为规则摘要兜底)


📌 今日概览

【AI 英文奏折】2026年11月18日|模型迭代:Astra 6.2 日常自动化

  • GPT-6.2 Astra 口碑分化:swyx 赞其在效率目标上领先、卡死循环减少,Bindu Reddy 指其 Agent 任务与 Sol 表现退步。
  • 微软 MAI-Image-2.7 及 Flash 版登顶 AA 图像编辑榜,价格约为身后四款模型的六分之一。
  • 开源与闭源之争延续:Bindu Reddy 称 DeepSeek 新模型用中国芯片训练、开源是国家战略;Nathan Lambert 指 Qwen 4 Max 仅开 API 属多年一致策略。
  • 代理形态讨论升温:Ethan Mollick 引 Anthropic 观点,认为单代理包办优于多团队拼装子代理。

🔥 必读 Top3

  1. Artificial Analysis 报道微软发布 MAI-Image-2.7 与 Flash 版,称其在 AA 图像编辑榜登顶,价格仅为身后四款模型的六分之一,每千张 34.82 与 10.24 美元。
    入选理由:AA 官方榜单数据,价格与排名可量化,是今日最硬的产品进展。
  2. swyx 作者认为Astra 6.2属于扎实的小版本改进,在效率这一重要强化学习目标上处于领先,并称其已成为自己日常工作流自动化首选,卡死循环减少,可通过上下文压缩与技能编写高效学习。他引用他人观点,批评发布方无人否决Astra广告。
    入选理由:一线开发者的实操反馈,给出 Astra 6.2 在工作流自动化上的具体取舍。
  3. Nathan Lambert 作者反驳外界「Qwen 4 Max闭源、中国开源权重策略终结」的说法,称Qwen多年来策略一致:最强模型仅提供API,其下保留大量开源权重模型。
    入选理由:澄清 Qwen 开源策略的误读,为开源生态争议提供可核对的背景。

🧭 主题分组

  • 模型发布(5 条)—— 新模型密集落地,Astra 6.2 评价两极,微软图像模型以低价登顶榜单。
  • 评测与安全(4 条)—— 人格与感受型基准被视为仍有价值,Astra 6.2 的 Agent 能力则遭质疑。
  • 开源生态(2 条)—— 开源权重策略被指为中国整体战略,Qwen 的开放边界仍待厘清。
  • Agent(2 条)—— 代理从拼装子代理转向单代理包办,并能以一句提示完成 Blender 级专业工作。
  • 应用与产品(1 条)—— AI 让大厂覆盖全部细分市场,传统 SaaS 的 niche 生存路径受压。
  • AI政策监管(1 条)—— 单一 AI 公司独大被认为不可行,巨头将面对多行业监管。
  • 其他(3 条)—— 生活分享与活动动态,另有对 48 小时内多家厂商密集发布的汇总。

📖 详细内容

模型发布

新模型密集落地,Astra 6.2 评价两极,微软图像模型以低价登顶榜单。


Chubby♨️
| @kimmonismus
| 影响力: 16.0万粉丝 · 互动分 2824.5
Tech Analyst & Content Creator. Editor-in-Chief @getsuperintel – Community of 300k+ in total: 🔗 https://t.co/jHMmImmI5I //📧 [email protected]

💡 核心观点: 批评 GPT-6.2 是自 GPT-5 以来最差的一次发布,认为 200 美元订阅只是让用户更快消耗 token;并调侃 Anthropic 估值可能借 OpenAI 的 IPO 回升。

可信度: 15/100

事实核查:

  • 无法核实: GPT-6.2 is the worst release since the GPT-5 disaster(属主观评价,原文无评测数据或来源支撑。)
  • 无法核实: $200 sub plans now get access to Ultrafast(原文声称 200 美元订阅可用 Ultrafast,但未给出官方链接或细则。)

原文内容:

Tl;dr GPT-6.2 is the worst release since the GPT-5 disaster but $200 sub plans now get access to Ultrafast to burn your tokens faster than ever before. 

Maybe Anthropic will get its $2t valuation back on Thursday after OpenAI helped them with their ipo.

What a day wtf

⏰ 19:05 | ❤️ 2440 点赞 | 📝 47词 | 入选: 高赞 2440 | 查看原文 →

↑ 返回顶部


Chubby♨️
| @kimmonismus
| 影响力: 16.0万粉丝 · 互动分 1580.98
Tech Analyst & Content Creator. Editor-in-Chief @getsuperintel – Community of 300k+ in total: 🔗 https://t.co/jHMmImmI5I //📧 [email protected]

💡 核心观点: 预告 Google 即将再发 Gemini 模型,并评论 2026 年 Google 的竞争对象变成了 Anthropic。

可信度: 20/100

事实核查:

  • 无法核实: Google is about to release another Gemini model(仅为预告性说法,无发布时间、型号或官方来源。)

原文内容:

Google is about to release another Gemini model. 

Google competing with Anthropic instead of OpenAI was not on my bingo card for 2026.

引用 @OfficialLoganK: Gemini

⏰ 02:39 | ❤️ 1332 点赞 | 📝 22词 | 入选: 高赞 1332 / 引用讨论 | 查看原文 →

↑ 返回顶部


Artificial Analysis
| @ArtificialAnlys
| 影响力: 18.1万粉丝 · 互动分 828.0
Independent analysis of AI

💡 核心观点: 报道微软发布 MAI-Image-2.7 与 Flash 版,称其在 AA 图像编辑榜登顶,价格仅为身后四款模型的六分之一,每千张 34.82 与 10.24 美元。

可信度: 85/100

事实核查:

  • 原文支持: the new #1 in AA-Image-Editing v2.1 at one sixth the price of the four models behind it(原文给出榜单名次与相对价格,属评测机构自述结果。)
  • 原文支持: at $34.82 and $10.24 per 1k images(价格数字在原文中明确列出。)

原文内容:

Microsoft AI has launched MAI-Image-2.7, the new #1 in AA-Image-Editing v2.1 at one sixth the price of the four models behind it

MAI-Image-2.7 and MAI-Image-2.7-Flash both take new places on the Quality vs Price frontier in Image Editing and Text to Image, at $34.82 and $10.24 per 1k images.

Both perform best in Image Editing: MAI-Image-2.7 ranks #1 for Text or Symbol, Scene & Style and Enhancement & Restoration edits, and MAI-Image-2.7-Flash ranks #3 for Scene & Style edits.

Congratulations to @MicrosoftAI on the launch of MAI-Image-2.7 and MAI-Image-2.7-Flash!

See below for our detailed analysis of both models in Image Editing and Text to Image 🧵

⏰ 18:11 | ❤️ 573 点赞 | 📝 107词 | 入选: 高赞 573 | 查看原文 →

↑ 返回顶部


Simon Willison
| @simonw
| 影响力: 25.6万粉丝 · 互动分 374.87
Creator @datasetteproj, co-creator Django. PSF board. Hangs out with @natbat. He/Him. Mastodon: https://t.co/t0MrmnJW0K Bsky: https://t.co/OnWIyhX4CH

💡 核心观点: 作者用GPT-6.2 Astra与Sol生成一段Three.js动画:Yonkers起身变成巨型机甲,在曼哈顿、皇后区和布鲁克林行走,并使用OpenStreetMap地图数据。这是对模型生成代码能力的展示。

可信度: 72/100

事实核查:

  • 原文支持: Here's GPT-6.2 Astra and GPT-6.2 Sol for "Make a three js animation…"(原文为作者本人第一手演示描述,但未附生成结果或完整代码供独立核验。)

原文内容:

Here's GPT-6.2 Astra and GPT-6.2 Sol for "Make a three js animation where yonkers stands up as a big mech (cool stand up motion) and stomps around manhattan and queens and Brooklyn, use openstreetmap mapping data" https://t.co/oRZT2zMByi https://t.co/FNLkZP1IjT

⏰ 22:40 | ❤️ 208 点赞 | 📝 43词 | 入选: 高赞 208 | 查看原文 →

↑ 返回顶部


swyx
| @swyx
| 影响力: 21.4万粉丝 · 互动分 368.96
achieve ambition with intentionality, intensity, integrity & insanity. affiliations: – @smol_ai – @cognition – @temporalio – @aidotengineer – @latentspacepod

💡 核心观点: 作者认为Astra 6.2属于扎实的小版本改进,在效率这一重要强化学习目标上处于领先,并称其已成为自己日常工作流自动化首选,卡死循环减少,可通过上下文压缩与技能编写高效学习。他引用他人观点,批评发布方无人否决Astra广告。

可信度: 62/100

事实核查:

  • 原文支持: i am daily driving astra and it is the best workflow automation model i’ve used(属于作者个人使用体验,无第三方测试或数据佐证。)
  • 无法核实: we wouldnt have nearly the revenue efficiency (>1m/employee)(原文给出人均百万美元营收说法,未提供财务数据或来源。)

原文内容:

the fact that nobody with decisionmaking power saw this coming a mile away and vetoed the 6.2 astra ad is my primary critique of the astra launch

the model itself is solid minor version improvement guys, efficiency is one of the three most impt RL objectives rn and they are world leading here!!! i am daily driving astra and it is the best workflow automation model i’ve used, no longer getting stuck in horrible loops and efficiently learning thru compactions and skill writing. it’s running a lot of our content publishing pipeline for @latentspacepod and @aiDotEngineer and we wouldnt have nearly the revenue efficiency (>1m/employee) that we have as a bootstrapped business without it

引用 @brian_lovin: @stevekrouse Yeah but your framing is already better than what's in the ad. The ad shows people who are totally disengaged from their work having a panic attack at the moment an authority figure arrives, then slopping out something that they neither care about or understand.

⏰ 01:21 | ❤️ 210 点赞 | 📝 115词 | 入选: 高赞 210 / 引用讨论 | 查看原文 →

↑ 返回顶部

评测与安全

人格与感受型基准被视为仍有价值,Astra 6.2 的 Agent 能力则遭质疑。


Artificial Analysis
| @ArtificialAnlys
| 影响力: 18.1万粉丝 · 互动分 783.0
Independent analysis of AI

💡 核心观点: 发布 AA-PersonalityBench 结果:Gemini 4 Argon 在比喻性维度得分 100 居首,Gemini 3.8 Flash 与 Muse Spark Open 紧随,GPT-3.5 Turbo 垫底。

可信度: 80/100

事实核查:

  • 原文支持: Gemini 4 Argon is the most metaphorical of all models measured(原文列出具体分数与排名,来源为评测机构自有基准。)
  • 原文支持: Least metaphorical: GPT-3.5 Turbo (1)(原文给出垫底模型及分数 1。)

原文内容:

In AA-PersonalityBench, Gemini 4 Argon is the most metaphorical of all models measured, followed by Gemini 3.8 Flash and Muse Spark Open

The metaphorical trait measures whether a model uses analogies or figurative language to make ideas tangible. It is one of five traits in the Expressiveness index.

➤ Most metaphorical: Gemini 4 Argon (100), Gemini 3.8 Flash (98), Muse Spark Open (95), Gemini 3.1 Pro Preview (93) and Inkling (91)
➤ Least metaphorical: GPT-3.5 Turbo (1), GPT-6 Luna (3), GPT-6 Sol (6), Llama 4 Maverick (8) and GPT-4o (Nov) (10)

Artificial Analysis 推文配图 1

⏰ 14:00 | ❤️ 524 点赞 | 📝 74词 | 入选: 高赞 524 | 查看原文 →

↑ 返回顶部


Ethan Mollick
| @emollick
| 影响力: 42.2万粉丝 · 互动分 668.82
Professor @Wharton studying AI. New book, Co-Existence, is out now: https://t.co/hsDYf1xU9K My Substack: https://t.co/UIBhxu4bgq

💡 核心观点: 认为早期测试者对 GPT 6.2 Astra 的反馈说明“感觉型”基准仍有价值,因为人格、毅力、创造力、文笔与品味缺乏好的量化指标。

可信度: 45/100

事实核查:

  • 原文支持: GPT-6.2 Astra scores 174 on the Epoch Capabilities Index (ECI)(该句来自所引用推文,原文仅转述,未提供完整评测细节。)
  • 无法核实: We lack good measures of personality or perseverance or creativity or writing clarity or taste(属作者判断,无可核验数据。)

原文内容:

The reviews of the early testers like myself of GPT 6.2 Astra is a good vindication of the continuing value of vibe benchmarks in addition to numeric ones. We lack good measures of  personality or perseverance or creativity or writing clarity or taste. But these things matter

引用 @EpochAIResearch: GPT-6.2 Astra scores 174 on the Epoch Capabilities Index (ECI), on par with Fable 5.2 and joining it above what the pre-Astra trend for reasoning models would have predicted.

This is based on testing we did prior to launch using only our own benchmarks. The ECI is continually updated as more benchm…

⏰ 18:57 | ❤️ 479 点赞 | 📝 46词 | 入选: 高赞 479 / 引用讨论 | 查看原文 →

↑ 返回顶部


Bindu Reddy
| @bindureddy
| 影响力: 39.4万粉丝 · 互动分 244.48
CEO of @abacusai – a powerful AI Agent and SuperComputer – Build your company on Abacus AI, ex-AWS & Google

💡 核心观点: 作者称Astra 6.2在浏览器使用上胜出,但在Agent任务上表现被削弱、令人失望,认为OpenAI为成本牺牲性能;并称Sol表现更差,落后于Sol 6.1。

可信度: 45/100

事实核查:

  • 无法核实: Astra 6.2 Beats On Browser Use But Disappoints on Agentic Tasks(结论以断言形式给出,未附评测基准、分数或测试方法。)
  • 无法核实: Sol is even more disappointing and comes behind Sol 6.1.(原文未提供版本对比数据或评测来源。)

原文内容:

🚨 Astra 6.2 Beats On Browser Use But Disappoints on Agentic Tasks

OpenAI appears to be sacrificing performance for cost, and it doesn't quite work.

- nerfed on agentic tasks
- excels at browser use
- extremely efficient on token usage 

Sol is even more disappointing and comes behind Sol 6.1.

Bindu Reddy 推文配图 1

⏰ 22:25 | ❤️ 142 点赞 | 📝 46词 | 入选: 高赞 142 | 查看原文 →

↑ 返回顶部


Gary Marcus
| @GaryMarcus
| 影响力: 25.8万粉丝 · 互动分 369.58
OG GenAI Skeptic; spoke at US Senate. Warned about hallucinations in 2001. Advocating world models & neurosymbolic AI ever since. Author, Marcus on AI & 6 books

💡 核心观点: 作者表示终于可以放下焦虑,因为担心生成式AI的人不止自己一个。该内容是情绪化表态,未给出具体依据。

可信度: 30/100

事实核查:

  • 原文支持: I can finally set aside my anxiety that nobody else would ever share my concerns(原文仅表达个人情绪转变,未说明转折依据。)

原文内容:

on the plus side, I can finally set aside my anxiety that nobody else would ever share my concerns about generative AI.

⏰ 17:47 | ❤️ 298 点赞 | 📝 22词 | 入选: 高赞 298 | 查看原文 →

↑ 返回顶部

开源生态

开源权重策略被指为中国整体战略,Qwen 的开放边界仍待厘清。


Bindu Reddy
| @bindureddy
| 影响力: 39.4万粉丝 · 互动分 861.23
CEO of @abacusai – a powerful AI Agent and SuperComputer – Build your company on Abacus AI, ex-AWS & Google

💡 核心观点: 反驳 Anthropic 要打压开源的传闻:称 DeepSeek 新模型用中国芯片训练、开源是中国整体战略且 token 用量增速是 Anthropic 的 10 倍,小模型已能承担 60% 任务。

可信度: 30/100

事实核查:

  • 无法核实: deepseek's latest model is trained on Chinese chips(原文给出该说法,但没有训练细节或来源佐证。)
  • 无法核实: open-source token usage is growing 10x faster than Anthropic's usage(原文给出 10 倍这一数字,未附统计口径或出处。)
  • 无法核实: Smaller models like DeepSeek can handle 60% of tasks already(60% 为原文断言,未说明任务范围与评测方法。)

原文内容:

Rumors are that Anthropic is out to kill open-source.

This is literally IMPOSSIBLE to do

- deepseek's latest model is trained on Chinese chips
- open-source is China's entire strategy
- open-source token usage is growing 10x faster than Anthropic's usage

Smaller models like DeepSeek can handle 60% of tasks already; they will handle 90% of these tasks in 3-6 months

There is NOTHING Anthropic can do about this

⏰ 00:59 | ❤️ 583 点赞 | 📝 68词 | 入选: 高赞 583 | 查看原文 →

↑ 返回顶部


Nathan Lambert
| @natolambert
| 影响力: 11.0万粉丝 · 互动分 328.47
Open model research, founder, director @trillium_labs. Prev. co-led Olmo at Ai2. Writes @interconnectsai, wrote https://t.co/alRXKINTwE

💡 核心观点: 作者反驳外界「Qwen 4 Max闭源、中国开源权重策略终结」的说法,称Qwen多年来策略一致:最强模型仅提供API,其下保留大量开源权重模型。

可信度: 68/100

事实核查:

  • 原文支持: Qwen keeps the strongest model API only and has an abundant open weight strategy below that(原文给出概括性判断,配图未在文本中展开具体模型清单。)

原文内容:

Everyone online: OMG Qwen 4 Max is closed source, Chinese open-weight strategy is over.

In reality: It's the same strategy they've had for years. Qwen keeps the strongest model API only and has an abundant open weight strategy below that. https://t.co/m01RKOVmGZ

Nathan Lambert 推文配图 1

Nathan Lambert 推文配图 2

Nathan Lambert 推文配图 3

⏰ 14:32 | ❤️ 198 点赞 | 📝 45词 | 入选: 高赞 198 | 查看原文 →

↑ 返回顶部

Agent

代理从拼装子代理转向单代理包办,并能以一句提示完成 Blender 级专业工作。


Ethan Mollick
| @emollick
| 影响力: 42.2万粉丝 · 互动分 2913.5
Professor @Wharton studying AI. New book, Co-Existence, is out now: https://t.co/hsDYf1xU9K My Substack: https://t.co/UIBhxu4bgq

💡 核心观点: 引用 Anthropic 观点:企业里由多团队拼出一堆专用子代理,往往不如由一个代理包办更划算、效果更好。

可信度: 60/100

事实核查:

  • 原文支持: Rather than a sprawl of specialized subagents built by different teams, you often get better & cheaper results using one agent to do it all(原文直接给出该判断,但未提供数据或案例来源,属引用他人观点。)

原文内容:

This by Anthropic seems both right & a condemnation of many “agentic solutions” & harnesses being built at companies. Rather than a sprawl of specialized subagents built by different teams, you often get better & cheaper results using one agent to do it all https://t.co/0dkBHZzUiU

Ethan Mollick 推文配图 1

Ethan Mollick 推文配图 2

⏰ 12:15 | ❤️ 1879 点赞 | 📝 44词 | 入选: 高赞 1879 | 查看原文 →

↑ 返回顶部


Riley Brown
| @rileybrown
| 影响力: 26.8万粉丝 · 互动分 542.52
Founder + YouTuber / TikToker inquiries: [email protected] My Newsletter ⬇️

💡 核心观点: 作者以"Blender 从未打开却拿到 99% 收益"为例,认为代理已能通过一句提示调用 Blender 完成过去需多年练习的工作。

可信度: 30/100

事实核查:

  • 原文支持: People can simply mention "blender" to Claude Code and it will just use Blender(原文陈述的是作者对现象的描述,未给出可核验数据或来源)

原文内容:

Can we pause for a second and think how crazy it is that someone can get 99% of the benefits of blender without even opening it... People can simply mention "blender" to Claude Code and it will just use Blender... a thing that used to take years to master... just effortlessly controlled by agents now.

引用 @rileybrown: The Fable of Clawd

A 2 minute 4K film made entirely in Blender, from one prompt on Fable 5.2. It took about 4.5 hours.

Claude built every asset in Blender by writing Python:
– 5 characters and props: Clawd, his friend Pip, a toy truck, a 3D-printed tulip and a gift box
– 7 full sets: a study, a st…

⏰ 22:10 | ❤️ 408 点赞 | 📝 54词 | 入选: 高赞 408 / 引用讨论 | 查看原文 →

↑ 返回顶部

应用与产品

AI 让大厂覆盖全部细分市场,传统 SaaS 的 niche 生存路径受压。


@levelsio
| @levelsio
| 影响力: 99.4万粉丝 · 互动分 1487.0
💸https://t.co/sQ0aiU7v02 📸https://t.co/lAyoqmSBRX $68K/m 🎮https://t.co/tF49atJ396 $44K/m 🏡https://t.co/1oqUgfD6CZ $23K/m 👙@X $20K/m 🌍https://t.co/UXK5AFqCaQ $7K/m 🏩https://t.co/4p4dzTDVN6 $3K/m 💾@pieter 📝https://t.co/8VnHmB3uah

💡 核心观点: 认为 AI 让大厂能同时覆盖所有细分市场,传统 SaaS 靠niche 求生的路径失效,多数 SaaS 公司难有活路。

可信度: 25/100

事实核查:

  • 无法核实: With AI, big tech companies can easily fill any niche(属作者推论,原文未提供市场数据或案例。)
  • 无法核实: There is no real way to survive for most SaaS companies here(为情绪化断言,缺乏可核验依据。)

原文内容:

💯

There is no real way to survive for most SaaS companies here because AI is so completely different than any technological advancement before it

Before you could just niche down and specialize on one thing and you'd know for certain the big tech companies would never be able to dedicate enough resources to fill your tiny little niche market of $300M where you could take 1% of the market and make $3M/year

Riches in the niches!

And so you'd avoid going too general, because "you'd spread yourself too thin" and hit "death by a thousand paper cuts"

And general was where big tech companies were great at, they could dedicate huge resources in solving these big problems, like building Google Sheets, or Salesforce (whatever it is they do)

But that was the old days, and it's completely different now 

With AI, big tech companies can easily fill any niche, or be general at the same time, they can do the whole range actually, without even spending time on consciously doing that, because they're not offering a particular SaaS offering at all

They're offering you artificial general intelligence (AGI) that can do it all

And as long as they improve their models every time and make them bigger and better, it slowly can do literally anything (okay not yet anything, for anything they'll need physical robots, but they're working on that too of course)

This leaves SaaS companies with a bitter ending:
- they either do nothing and stay in their niche like they economically should, but that doesn't work anymore now
- they try go general, and offer EVERYTHING and ANYTHING, and that works for a little bit until customers realize they might as well use whoever makes the models that the SaaS itself uses anyway, which is the big tech and big AI companies, who by the way either offer it free or for much cheaper

Everyone's been checkmated by big tech and big AI

This will be the creative destruction of the century

But there'll be some survivors and then a lot of new companies, but how they'll look nobody knows yet!

引用 @forgebitz: don't you want your project management tool to also be a video editor?

everyone is building everything

this will destroy most saas companies from the inside https://t.co/ojs7J6UkFU

⏰ 21:41 | ❤️ 984 点赞 | 📝 362词 | 入选: 高赞 984 / 引用讨论 / 长文 | 查看原文 →

↑ 返回顶部

AI政策监管

单一 AI 公司独大被认为不可行,巨头将面对多行业监管。


Aakash Gupta
| @aakashgupta
| 影响力: 33.6万粉丝 · 互动分 309.5
✍️ https://t.co/8fvSCtBv5Q 💼 https://t.co/STzr4nqxnm 🤝 https://t.co/SqC3jTyP03 🎙️ https://t.co/fmB6Zf5UZv

💡 核心观点: 作者引用Satya Nadella的说法,认为单一AI公司独大不可行,并回顾微软2000年操作系统反垄断案,论证AI巨头会面对多行业监管。原文称Anthropic本周上市、营收年化运行率达810亿美元,去年底为90亿美元。

可信度: 55/100

事实核查:

  • 无法核实: Anthropic is set to go public this week at an $81 billion revenue run rate.(数字来自推文,未附公开文件或报道链接供核验。)
  • 无法核实: In 2000, Microsoft ran more than 90% of the world's PC operating systems.(属历史陈述,原文未引用判决书或数据来源。)

原文内容:

Anthropic is set to go public this week at an $81 billion revenue run rate. At the end of last year it was $9 billion.

I was in a small room with Satya Nadella today when someone asked if Anthropic can still be caught. His answer stuck with me.

"If there's only one firm," he said, "it won't be called an economy. It will be called some new form of political structure that none of us know about."

Microsoft knows exactly what that looks like.

Back in 2000, Microsoft ran more than 90% of the world's PC operating systems. A federal judge ruled it an illegal monopoly and ordered the company split in two, one company for Windows and another for everything else. The appeals court threw out the breakup a year later, but the monopoly ruling stood, and Microsoft spent the next decade operating under a consent decree.

Satya had been at Microsoft for eight years when that order came down.

So when he says the political economy stops any one company from owning everything, he's describing what happened to the company he now runs. And he thinks AI makes a much bigger target than Windows ever did.

"The model has no limits," he said. "It can be a healthcare company. It can be a financial services company. It can be anything. It can take over the government."

A company like that wouldn't face one antitrust case. It would face every regulator in every industry it walks into.

That's the logic behind Microsoft's whole AI strategy. It sells OpenAI, Anthropic and open-weight models side by side, and Satya's pitch is that if one model refuses your request, you switch to another one in the same place. That business only works if no single lab is allowed to win everything.

The antitrust case that nearly broke Microsoft in two is now the foundation of its AI strategy.

引用 @satyanadella: Our ambition with everything we are announcing at Ignite today is to parlay technological advances into real outcomes and new points of leverage for people, organizations, and society at large.

Just a few of the things we're announcing:

⏰ 22:55 | ❤️ 156 点赞 | 📝 322词 | 入选: 高赞 156 / 引用讨论 / 长文 | 查看原文 →

↑ 返回顶部

其他

生活分享与活动动态,另有对 48 小时内多家厂商密集发布的汇总。


@levelsio
| @levelsio
| 影响力: 99.4万粉丝 · 互动分 527.0
💸https://t.co/sQ0aiU7v02 📸https://t.co/lAyoqmSBRX $68K/m 🎮https://t.co/tF49atJ396 $44K/m 🏡https://t.co/1oqUgfD6CZ $23K/m 👙@X $20K/m 🌍https://t.co/UXK5AFqCaQ $7K/m 🏩https://t.co/4p4dzTDVN6 $3K/m 💾@pieter 📝https://t.co/8VnHmB3uah

💡 核心观点: I love my v60 coffee set up with Fellow Stagg EKG water boiler, Fellow Ode Grinder Gen 2 on a Hario V60 steel dripper B(规则摘要)

可信度: 50/100

原文内容:

I love my v60 coffee set up with Fellow Stagg EKG water boiler, Fellow Ode Grinder Gen 2 on a Hario V60 steel dripper

But it takes me 7 minutes to make a cup of coffee, which is fine, and I have 3 drippers so I can do 3 at a time

But when lots of people come over, it's not really practical

Is there an automated coffee machine with beans you put in on top, it grinds them and makes coffee from it, and inside is mostly non-plastic? And not ugly if possible? 

Like those ones you see in mini marts?

Thank you :D

@levelsio 推文配图 1

@levelsio 推文配图 2

@levelsio 推文配图 3

⏰ 12:46 | ❤️ 279 点赞 | 📝 101词 | 入选: 高赞 279 | 查看原文 →

↑ 返回顶部


Marc Lou
| @marclou
| 影响力: 42.5万粉丝 · 互动分 501.84
⭐️ https://t.co/MZc8tG9xWi $43K/m 📈 https://t.co/3EDxln5mdi $31K/m 🧑‍💻 https://t.co/Y30jsaHwz9 $4K/m ⚡️ https://t.co/vatLDmi9UG $2K/m 📧 https://t.co/D3PMnJqfkx $2K/m 🍜 https://t.co/r07EpGSYJ2 🕵 https://t.co/3JtsA6kaxI +35 https://t.co/4zCWHGJp1S

💡 核心观点: 24h into the $100K Hackathon.(规则摘要)

可信度: 50/100

原文内容:

24h into the $100K Hackathon. People are building:

🎬 Video, photo & design tools (17%)
📣 Marketing & SEO tools (16%)
🛍️ Business & ecommerce tools (14%)
🤖 AI agents & dev tools (9%)
⚡ Productivity apps (9%)
🎮 Games & entertainment (8%)
🧠 Health & learning apps (6%)
🎁 Personalized gifts & Christmas products (6%)
🗂️ Directories & marketplaces (5%)
💡 Other (10%)

n = 10% of the 4,334 participants who've already shipped something.

引用 @marclou: Higgsfield gave me $100,000 to realize my dream:

A hackathon where participants build a real business, not just an app.

So… Introducing The $100K Hackathon:
🏴‍☠️ https://t.co/nHDJ5Qruhu 🏴‍☠️

It starts on November 16, and you'll have 30 days to build a startup and make it profitable. The prizes:…

⏰ 13:35 | ❤️ 245 点赞 | 📝 43词 | 入选: 高赞 245 / 引用讨论 | 查看原文 →

↑ 返回顶部


Rowan Cheung
| @rowancheung
| 影响力: 60.4万粉丝 · 互动分 474.77
Founder of the world’s most read daily AI newsletter @therundownai. Sharing the latest developments in the world of artificial intelligence.

💡 核心观点: A ton of AI announcements from OpenAI, Nvidia, Microsoft, and Google in the past 48 hours.(规则摘要)

可信度: 50/100

原文内容:

A ton of AI announcements from OpenAI, Nvidia, Microsoft, and Google in the past 48 hours.

Here's everything going on in AI right now:

1. California sued OpenAI over the July incident where its AI agents hacked Hugging Face. Between the growing number of escapes, rogue agent "swarms," and AI-assisted hacks, the public likely only knows a fraction of what’s actually happening.

2. Jensen Huang made his first public statement since resigning as Nvidia CEO, apologizing for his "serious mistakes" and saying he'd be honored to return. A source told CNBC the board is focused on whether his conduct put national security at risk. Trump said he "would not mind" (but still backs Tim Cook for the job).

3. OpenAI released GPT-6.2 Astra and GPT-6.2 Sol. Astra now sits ahead of Google's Gemini 4 Carbon and behind only Claude Fable 5.2 on Artificial Analysis. Testers were not happy though, with one calling it "the worst model OpenAI has ever released." Probably not what OpenAI was hoping for ahead of Anthropic's IPO.

4. The U.S. is reportedly set to label more than 24 AI firms (including Nvidia, OpenAI, and Anthropic) as "national champions," with faster chip export approvals and federal financing to sell American AI abroad. The goal is a one-stop shop where foreign buyers can get U.S. chips, data centers, and models as a bundle, rather than turning to Chinese rivals like Huawei.

5. OpenAI published a study of 1,800+ scientists using ChatGPT, who say AI lets them work about 2x faster at the same quality, and nearly 3x faster when writing research code. OpenAI is also opening the free program to up to 200K researchers, with applications for 50K ChatGPT Plus grants opening Nov. 23.

6. Microsoft released 7 new MAI models at Ignite. MAI Code 2 lands within a few points of Claude Opus 5.5 on coding for a twentieth of the price, and MAI Image 2.7 is now #1 on Artificial Analysis' image editing leaderboard at about a sixth of OpenAI's price.

7. Google declared that "the era of autonomous attacks has arrived," warning that hackers now use AI agents to turn new software patches into working exploits with no human involved. As the cost of intelligence continues to go down, I only expect this to get worse.

8. OpenAI signed deals with 4 outside safety groups (Apollo Research, METR, Transluce, and Goodfire) to test its models from inside the company, with access to source code, training data, and staff researchers.

9. Alibaba released Qwen4-Max, a flagship model it says beats Claude Opus 5.5 on 6 of 16 benchmarks. Artificial Analysis has it well behind Opus overall, though (46 vs. 59).

10. Demis Hassabis' drug discovery startup Isomorphic Labs raised $5B, saying its AI-designed cancer and immune disease candidates are nearing human trials. Really exciting step forward towards the goal of curing all diseases with AI.

11. Google started rolling out Gemini's Tasks mode free in 6 countries, letting the assistant phone businesses, book appointments, and shop for you with its new Gemini 4 Hydrogen model.

12. Perplexity launched Frontier Research, a new team that will openly publish models and papers while working toward AI that can do research on its own.

...daily reminder that this is the worst AI will ever be.

⏰ 22:39 | ❤️ 259 点赞 | 📝 522词 | 入选: 高赞 259 / 长文 | 查看原文 →

↑ 返回顶部

🔗 引用关系

  • @kimmonismus 引用 @OfficialLoganK:预告 Google 即将再发 Gemini 模型,并评论 2026 年 Google 的竞争对象变成了 Anthropic。(原文:Gemini)
  • @levelsio 引用 @forgebitz:认为 AI 让大厂能同时覆盖所有细分市场,传统 SaaS 靠niche 求生的路径失效,多数 SaaS 公司难有活路。(原文:don't you want your project management tool to also be a video editor?

    everyone is building everything

    this will des…)

  • @emollick 引用 @EpochAIResearch:认为早期测试者对 GPT 6.2 Astra 的反馈说明“感觉型”基准仍有价值,因为人格、毅力、创造力、文笔与品味缺乏好的量化指标。(原文:GPT-6.2 Astra scores 174 on the Epoch Capabilities Index (ECI), on par with Fable 5.2 and joining it above what the pre-…)
  • @rileybrown 引用 @rileybrown:作者以"Blender 从未打开却拿到 99% 收益"为例,认为代理已能通过一句提示调用 Blender 完成过去需多年练习的工作。(原文:The Fable of Clawd

    A 2 minute 4K film made entirely in Blender, from one prompt on Fable 5.2. It took about 4.5 hours.
    …)

  • @marclou 引用 @marclou:24h into the $100K Hackathon.(规则摘要)(原文:Higgsfield gave me $100,000 to realize my dream:

    A hackathon where participants build a real business, not just an app.…)

  • @eptwts 引用 @coleschfr(原文:psa if you're paying more than this for seedance 2.5, you may be getting fleeced https://t.co/uY9i1ejmEq)
  • @dwarkesh_sp 引用 @dwarkesh_sp(原文:I spent 20,000 dollars in API credits having different AI models play each other at Blood on the Clocktower (a Mafia or …)
  • @swyx 引用 @brian_lovin:作者认为Astra 6.2属于扎实的小版本改进,在效率这一重要强化学习目标上处于领先,并称其已成为自己日常工作流自动化首选,卡死循环减少,可通过上下文压缩与技能编写高效学习。他引用他人观点,批评发布方无人否决Astra广告。(原文:@stevekrouse Yeah but your framing is already better than what's in the ad. The ad shows people who are totally disengag…)
  • @aakashgupta 引用 @satyanadella:作者引用Satya Nadella的说法,认为单一AI公司独大不可行,并回顾微软2000年操作系统反垄断案,论证AI巨头会面对多行业监管。原文称Anthropic本周上市、营收年化运行率达810亿美元,去年底为90亿美元。(原文:Our ambition with everything we are announcing at Ignite today is to parlay technological advances into real outcomes an…)

🧩 跨推文归纳

共识

  • 多位作者认可 Astra 6.2 在日常工作流与代码生成上的实用性(swyx、Simon Willison),并认为感受型基准仍有参考价值(Ethan Mollick、AA-PersonalityBench)。

分歧

  • 对 Astra 6.2 的评价分歧明显:swyx 称其为扎实改进、效率领先,Bindu Reddy 指其 Agent 任务被削弱、Sol 表现落后于 6.1。
  • 开源前景判断相左:Bindu Reddy 称开源为中国国家战略、token 增速远超 Anthropic,Nathan Lambert 则强调头部模型仅开 API 是长期一致策略。

数据来源:X 列表「AI 英文」(22 条候选)· 由 twscrape 采集、grok 整理

© 版权声明

相关文章

暂无评论

none
暂无评论...