共收录 18 篇深度内容（其中 6 条为规则摘要兜底）
📌 今日概览
- Gemini 与 DeepSeek 同时把能力推向可直接使用的场景：前者演示从档案素材剪出配解说成片，后者把 App 的搜索、文件与语音开放成 API。
- 评测口径连续更新，编码代理与图像编辑榜单同日刷新，开源权重模型在细分榜单位次抬升。
- decision model 成为多家厂商一个多月内集中推出的新品类，有第三方数据显示 26% 请求已走该路由；算力侧则出现 kernel 优化带来的芯片对比变化。
- 产品与商业侧讨论集中在成本核算与垂直行业 agent 的工具栈搭建。
🔥 必读 Top3
- Google Gemini Google Gemini 官方账号展示了用 Gemini 4 Hydrogen 的 custom skill，从公开领域档案影像中采集素材、剪辑成一条连续成片并配解说，把想法直接做成一段宣传视频。
入选理由：官方演示从公开档案影像取材到配解说成片的完整链路，是今日落地最具体的生成式视频案例。 - Artificial Analysis Artificial Analysis 发布 Coding Agent Index v2.1，升级 FrontierCode 与 Terminal-Bench，Claude Fable 5.2 与 Claude Opus 5.5 以 58 分并列领先。
入选理由：Coding Agent Index 升到 v2.1 并给出 Claude Fable 5.2 与 Claude Opus 5.5 并列领先，直接影响编码代理选型。 - DeepSeek 🚀 The full DeepSeek App experience, now via API Search, files, links, safety, and voice.（规则摘要）
入选理由：DeepSeek App 的搜索、文件、链接、安全与语音能力开放为 API，是开发者最关心的接入面变化。
🧭 主题分组
- 应用与产品（5 条）—— 官方与独立开发者都在验证两类路径：AI 直接产出成片，以及为垂直行业搭建 agent 的生意。
- 评测与榜单（4 条）—— 编码代理与图像编辑榜单同日更新，闭源模型并列领先，开源权重模型在细分榜单位次抬升。
- 模型与训练（2 条）—— decision model 成为多家厂商密集推出的新品类，模型命名体系同时出现混乱。
- 算力与芯片（1 条）—— kernel 层优化让 MI355X 上的 vLLM 在部分交互场景取得对 Blackwell 的领先。
- AI 政策（1 条）—— 芯片出口受审查的同时，向中国出售数据仍处规则空白，被指构成蒸馏路径。
- AI 视频（1 条）—— MiniMax H3 以限时免费开放无限使用，主打局部编辑角色、动作与声音。
- 行业动向（4 条）—— 数学领域的冲击被视为即将传导到 AI 研发本身，安全评估与厂商对话同步推进。
📖 详细内容
应用与产品
官方与独立开发者都在验证两类路径：AI 直接产出成片，以及为垂直行业搭建 agent 的生意。
Google Gemini
｜ @GeminiApp
｜ 影响力: 61.6万粉丝 · 互动分 1708.0
Google Gemini
｜ @GeminiApp
｜ 影响力: 61.6万粉丝 · 互动分 1708.0
💡 核心观点: Google Gemini 官方账号展示了用 Gemini 4 Hydrogen 的 custom skill，从公开领域档案影像中采集素材、剪辑成一条连续成片并配解说，把想法直接做成一段宣传视频。
可信度: 55/100
事实核查:
- 无法核实: Gemini 4 Hydrogen created a sizzle reel about the element “Hydrogen,” sourcing public-domain archival footage（原文自述，未展示成片或素材来源清单；属于官方产品演示，细节无法独立核实）
原文内容:
Go from idea to finished video with Gemini 4 Hydrogen. Using a custom skill and simple brief, Gemini 4 Hydrogen created a sizzle reel about the element “Hydrogen," sourcing public-domain archival footage and cutting it into one continuous edit with clean narration.
⏰ 22:05 | ❤️ 1245 点赞 | 📝 41词 | 入选: 高赞 1245 | 查看原文 →
Marc Lou
｜ @marclou
｜ 影响力: 42.4万粉丝 · 互动分 1601.5
Marc Lou
｜ @marclou
｜ 影响力: 42.4万粉丝 · 互动分 1601.5
💡 核心观点: 作者称用 Opus 5.5 在 Cursor 上开发 MailCheap 花了约 4000 美元，因此这款产品成了他第一个不赚钱的产品。
可信度: 40/100
事实核查:
- 无法核实: It cost me ~$4,000 to build MailCheap with Opus 5.5（原文给出金额与所用模型工具，但无账单或成本明细佐证）
- 无法核实: it's my first unprofitable product（作者自述经营结果，原文未提供收入数据）
原文内容:
It cost me ~$4,000 to build MailCheap with Opus 5.5 on @cursor_ai so ironically, it's my first unprofitable product. https://t.co/cQn0TETUYA
⏰ 13:15 | ❤️ 911 点赞 | 📝 20词 | 入选: 高赞 911 | 查看原文 →
GREG ISENBERG
｜ @gregisenberg
｜ 影响力: 76.3万粉丝 · 互动分 1183.5
GREG ISENBERG
｜ @gregisenberg
｜ 影响力: 76.3万粉丝 · 互动分 1183.5
💡 核心观点: 作者认为当前最好的生意之一是为某个垂直行业（如牙医）搭建 Grok Bot 和 AI agent，并列出了 T3 Code、Paper、Orgo、AgentMail、AgentPhone、Composio 等工具栈。
可信度: 25/100
事实核查:
- 无法核实: One of the best businesses to build right now: a Grok Bot for one industry.（属个人判断与创业建议，原文无市场数据或案例支撑）
- 原文支持: The stack: – T3 Code + Opus 5.5/Fable 5.2 to build it – Paper to design every screen first – Orgo to give each agent its own computer（工具名均见于原文，但各工具的实际用途与适配性未验证）
原文内容:
One of the best businesses to build right now: a Grok Bot for one industry. Pick a niche like dentists and build the AI agents for them so they don’t have to figure any of it out. The stack: - T3 Code + Opus 5.5/Fable 5.2 to build it - Paper to design every screen first - Orgo to give each agent its own computer - AgentMail for an inbox - AgentPhone for a phone number - Composio for every connector - CleanShot to screenshot bugs and say “fix this” Dentists fix teeth and you set up the agents. Really good way to learn how to become AI native while getting paid for it too.
⏰ 15:07 | ❤️ 680 点赞 | 📝 106词 | 入选: 高赞 680 | 查看原文 →
DeepSeek
｜ @deepseek_ai
｜ 影响力: 121.0万粉丝 · 互动分 10756.5
DeepSeek
｜ @deepseek_ai
｜ 影响力: 121.0万粉丝 · 互动分 10756.5
💡 核心观点: 🚀 The full DeepSeek App experience, now via API Search, files, links, safety, and voice.（规则摘要）
可信度: 50/100
原文内容:
🚀 The full DeepSeek App experience, now via API Search, files, links, safety, and voice. The capabilities behind the DeepSeek App are now available through the DeepSeek API. Enable built-in features with a few parameters, or call standalone APIs such as OCR, ASR, and TTS. Build richer conversations or recreate the full DeepSeek App in your own application. Try our open-source frontend demo, with support for importing DeepSeek conversation history. 🔗 Demo: https://t.co/TI0Iw5qLQK 📦 Code: https://t.co/ImA0LrgUgX 📖 Details: https://t.co/t0dJxXM6Lk 1/4
⏰ 14:24 | ❤️ 8024 点赞 | 📝 83词 | 入选: 高赞 8024 | 查看原文 →
ChatGPT
｜ @ChatGPT
｜ 影响力: 79.1万粉丝 · 互动分 6247.5
ChatGPT
｜ @ChatGPT
｜ 影响力: 79.1万粉丝 · 互动分 6247.5
💡 核心观点: Now your dot can securely complete purchases on your behalf—with your approval.（规则摘要）
可信度: 50/100
原文内容:
Now your dot can securely complete purchases on your behalf—with your approval. With ChatGPT Wallet, agents pay without access to your actual card details. Every purchase requires a security check. ChatGPT Wallet is rolling out starting today to dots, for Pro subscribers in the US. https://t.co/OnAl3SKQ3n
⏰ 16:44 | ❤️ 3962 点赞 | 📝 49词 | 入选: 高赞 3962 | 查看原文 →
评测与榜单
编码代理与图像编辑榜单同日更新，闭源模型并列领先，开源权重模型在细分榜单位次抬升。
Artificial Analysis
｜ @ArtificialAnlys
｜ 影响力: 17.9万粉丝 · 互动分 801.5
Artificial Analysis
｜ @ArtificialAnlys
｜ 影响力: 17.9万粉丝 · 互动分 801.5
💡 核心观点: Artificial Analysis 发布 Coding Agent Index v2.1，升级 FrontierCode 与 Terminal-Bench，Claude Fable 5.2 与 Claude Opus 5.5 以 58 分并列领先。
可信度: 85/100
事实核查:
- 原文支持: Claude Fable 5.2 and Claude Opus 5.5 share the lead at 58（原文明确给出分数与并列结论，属机构自建榜单结果）
- 原文支持: v2.1 of the Index upgrades FrontierCode from 1.1 to 2.0 and Terminal-Bench from 4.0 to 4.1.（原文陈述版本升级内容，与正文一致）
- 原文支持: models score lower by 5 points on average（原文给出平均降幅，但未披露完整样本与统计方法）
原文内容:
Announcing Artificial Analysis Coding Agent Index v2.1, upgrading FrontierCode and Terminal-Bench. Claude Fable 5.2 and Claude Opus 5.5 share the lead at 58 The Coding Agent Index measures models in agent harnesses; v2.1 of the Index upgrades FrontierCode from 1.1 to 2.0 and Terminal-Bench from 4.0 to 4.1. Upgraded: ➤ FrontierCode-AA 2.0 (replacing FrontierCode v1.1): our implementation of Cognition's FrontierCode 2.0, comprising 132 real-world coding tasks graded on whether a repository maintainer would merge the agent's change ➤ Terminal-Bench 4.1 (replacing Terminal-Bench 4.0): the same 66 tasks, scored with updated verifiers that improve reliability and resistance to reward hacking Staying in the Index: SWE-Atlas-QnA-AA Key results: ➤ Claude Fable 5.2 and Claude Opus 5.5 share the lead at 58: Claude Fable 5.2 scores 58 at both xhigh and high in Claude Code. Claude Opus 5.5 (max) scores 58 in Claude Code and, paired with GPT-6 Luna High, in Cognition's multi-model harness Devin Fusion CLI. Claude Fable 5.2 (max) scores 56 as more of its attempts hit task time limits. Claude Sonnet 5.5 (max, 56) in Claude Code and Gemini 4 Carbon (high, 55) in Antigravity CLI follow ➤ Index scores are lower in v2.1, driven by the upgrade to FrontierCode-AA 2.0: models score lower by 5 points on average. Codex with GPT-6.1 Terra (max) is the only configuration which didn’t fall between versions, sitting within 1 point of its prior score ➤ Cost per task varies ~30x among configurations scoring 48+: from $1.51 for Codex with GPT-6.1 Sol (max, 52) to $44.22 for Claude Code with Claude Fable 5.2 (max, 56)
⏰ 00:04 | ❤️ 483 点赞 | 📝 231词 | 入选: 高赞 483 / 长文 | 查看原文 →
Artificial Analysis
｜ @ArtificialAnlys
｜ 影响力: 17.9万粉丝 · 互动分 346.24
Artificial Analysis
｜ @ArtificialAnlys
｜ 影响力: 17.9万粉丝 · 互动分 346.24
💡 核心观点: Artificial Analysis 称 MiniMax H3-Image 成为 AA-Image-Editing v2.1 与 AA-Image-T2I v2.1 上排名第一的开源权重图像模型，总榜分别列第 13 与第 25。
可信度: 80/100
事实核查:
- 原文支持: MiniMax H3-Image is the new #1 open weights image model on both AA-Image-Editing v2.1 and AA-Image-T2I v2.1（原文明确给出双榜第一结论）
- 原文支持: Overall, H3-Image ranks #13 in Image Editing and #25 in Text to Image.（原文给出总榜位次，与开源权重排名区分清晰）
原文内容:
MiniMax H3-Image is the new #1 open weights image model on both AA-Image-Editing v2.1 and AA-Image-T2I v2.1 Overall, H3-Image ranks #13 in Image Editing and #25 in Text to Image. H3-Image Highspeed, a faster version on MiniMax's API only, ranks #14 and #24. Congratulations to @MiniMax_AI on the release! See below for our analysis of both models in the Artificial Analysis Image Arena 🧵
⏰ 20:29 | ❤️ 248 点赞 | 📝 59词 | 入选: 高赞 248 | 查看原文 →
Santiago
｜ @svpino
｜ 影响力: 45.7万粉丝 · 互动分 276.99
Santiago
｜ @svpino
｜ 影响力: 45.7万粉丝 · 互动分 276.99
💡 核心观点: Deepnote 发布 Agent Initiative Benchmark，在模拟公司的 Slack 中测试 12 个前沿模型；代理可访问 11 项服务的记录，面对 50 种情境，须在 3 分钟内判断是否介入、如何帮忙、何时保持沉默。
可信度: 70/100
事实核查:
- 原文支持: It's a benchmark released by Deepnote.（推文明确标注发布方与基准名称。）
- 原文支持: They face 50 different situations（50 种情境、11 项服务、3 分钟均见于原文。）
- 无法核实: We tested 12 frontier models（被引用推文称测试 12 个前沿模型，未在原文给出具体成绩。）
原文内容:
This is a really good test for agents: It's a benchmark released by Deepnote. (I've used their platform for 5+ years and I love it!) The benchmark tests agents in a simulated company's Slack: • Agents get access to records across 11 services • They face 50 different situations • They have 3 minutes to investigate and decide Here is what they test: 1. Does the agent know when to help? 2. Does the agent know how to help accurately? 3. Does the agent know when to stay quiet?
引用 @jakubjurovych: What’s the best model for proactive agents?
Today, we’re introducing the Agent Initiative Benchmark. We tested 12 frontier models on realistic workplace tasks inside a synthetic company with 6 years of highly detailed history: it has a financial model, CRM records, past-due invoices, Linear tickets…
⏰ 15:28 | ❤️ 135 点赞 | 📝 82词 | 入选: 高赞 135 / 引用讨论 | 查看原文 →
Charly Wargnier ♨️
｜ @DataChaz
｜ 影响力: 18.3万粉丝 · 互动分 225.01
Charly Wargnier ♨️
｜ @DataChaz
｜ 影响力: 18.3万粉丝 · 互动分 225.01
💡 核心观点: Reason（reasonmachines，YC F26）发布编码代理 harness，宣称成本最多降低 20 倍，并在 Terminal-Bench 与 DeepSWE 上胜过顶级 harness；提供 Slack 集成、API 与 macOS 应用，可与 Claude Code、Codex、Cursor 并行运行。
可信度: 40/100
事实核查:
- 无法核实: up to 20x lower cost（成本与基准成绩均为发布方自述，原文无第三方验证。）
- 无法核实: beats top harnesses on Terminal-Bench and DeepSWE, per their numbers（作者已注明数据来自发布方，未见独立复现。）
原文内容:
Reason from @reasonmachines (YC F26) just launched. and it puts so many useful pieces together 🔥 > up to 20x lower cost > beats top harnesses on Terminal-Bench and DeepSWE, per their numbers > Slack integration for getting work back in the thread > API, macOS app > runs beside Claude Code, Codex and Cursor Lower agent bills, with many practical ways to put the agents to work ↓
引用 @xadisingh: Frontier coding performance at up to 20x lower cost.
Introducing the world’s most cost-efficient harness: Reason
Stop burning thousands on runaway agent bills.
Reason outperforms top harnesses on Terminal-Bench and DeepSWE while dropping cost per task to a fraction of the frontier baseline.
Com…
⏰ 16:13 | ❤️ 112 点赞 | 📝 61词 | 入选: 高赞 112 / 引用讨论 | 查看原文 →
模型与训练
decision model 成为多家厂商密集推出的新品类，模型命名体系同时出现混乱。
Sebastian Raschka
｜ @rasbt
｜ 影响力: 53.8万粉丝 · 互动分 733.53
Sebastian Raschka
｜ @rasbt
｜ 影响力: 53.8万粉丝 · 互动分 733.53
💡 核心观点: 作者观察到各大厂商在一个多月内密集推出各自的 decision model，并引用 Vercel 报告称 26% 的请求已通过 decision model 路由。
可信度: 60/100
事实核查:
- 原文支持: now pretty much everyone launched their own decision model copycats（原文列出 15 家机构与对应发布时间，可按列表逐条比对）
- 无法核实: According to a Vercel report, 26% of requests are now routed through decision models（转述第三方报告数据，原文未给出报告链接或口径）
原文内容:
Wow that escalated quickly. Was gone for a few weeks and now pretty much everyone launched their own decision model copycats: - Sep 23: Together AI, Tev1-4B - Sep 24: Alibaba Cloud, decision-model - Sep 29: Liquid AI, d1 - Sep 30: Databricks, ai_decide - Oct 1: Amazon AWS, Strands Decider 2B - Oct 1: Cloudflare, Clef and Clef-flash (Clef-omni on Oct 9) - Oct 1: Perplexity, Decider (v1.3 on Nov 12) - Oct 6: OpenAI, Decisions API (GPT-6 Luna) - Oct 7: https://t.co/s8O5rwbq5z, H2O-Lightning-4B - Oct 7: Snowflake, Snowflake Decision - Oct 9: Microsoft, Microsoft-Decision-1 - Oct 15: IBM, Granite Switch 4.1 with Decision adapter - Oct 21: Baseten, Mammoth-1 and Elephant-1 - Nov 6: Salesforce, Koa Decisions - Nov 11: Google DeepMind, Gemini Decision 1 According to a Vercel report, 26% of requests are now routed through decision models 📈
引用 @rasbt: Took a while, but here it is, a mega write-up on language models for text classification and, of course, Jev.
It's basically a visual guide to RNNs, CNNs, transformers, and calibration, with hands-on experiments on accuracy and efficiency. https://t.co/fVHpuKvu5o
⏰ 19:12 | ❤️ 512 点赞 | 📝 107词 | 入选: 高赞 512 / 引用讨论 | 查看原文 →
Ethan Mollick
｜ @emollick
｜ 影响力: 41.8万粉丝 · 互动分 371.9
Ethan Mollick
｜ @emollick
｜ 影响力: 41.8万粉丝 · 互动分 371.9
💡 核心观点: 作者用一段调侃对话呈现模型命名混乱：hydrogen 与 sonnet、luna、spark 同级，fable 则对应 wonder、astra、carbon（原本类似 argon）。
可信度: 15/100
原文内容:
"You see the model called hydrogen is equivalent to the models called sonnet and luna and spark, but the model called fable is like wonder and astra and carbon (but it was like argon)" "Wait, why do you look confused?"
⏰ 20:17 | ❤️ 207 点赞 | 📝 40词 | 入选: 高赞 207 | 查看原文 →
算力与芯片
kernel 层优化让 MI355X 上的 vLLM 在部分交互场景取得对 Blackwell 的领先。
SemiAnalysis
｜ @SemiAnalysis_
｜ 影响力: 18.6万粉丝 · 互动分 593.82
SemiAnalysis
｜ @SemiAnalysis_
｜ 影响力: 18.6万粉丝 · 互动分 593.82
💡 核心观点: SemiAnalysis 称 Inferact 团队用 FlyDSL kernel 让 MI355X 上的 vLLM 在 MiniMax M3.1 上跑赢 Blackwell，部分交互场景甚至超过 Grace Blackwell。
可信度: 55/100
事实核查:
- 无法核实: MI355X vLLM just beat Blackwell on MiniMax M3.1, even surpassing Grace Blackwell at some interactivities（原文为系列推文首条，未给出基准配置与完整数据）
- 原文支持: By leveraging FlyDSL kernels, the @Inferact team brought the full potential out from the MI355X GPU（技术手段在原文中明确，效果归因属作者判断）
原文内容:
Cracked kernel alert 🚨 MI355X vLLM just beat Blackwell on MiniMax M3.1, even surpassing Grace Blackwell at some interactivities. By leveraging FlyDSL kernels, the @Inferact team brought the full potential out from the MI355X GPU (1/3) https://t.co/R57VQmrPsF
⏰ 16:06 | ❤️ 421 点赞 | 📝 34词 | 入选: 高赞 421 | 查看原文 →
AI 政策
芯片出口受审查的同时，向中国出售数据仍处规则空白，被指构成蒸馏路径。
SemiAnalysis
｜ @SemiAnalysis_
｜ 影响力: 18.6万粉丝 · 互动分 285.0
SemiAnalysis
｜ @SemiAnalysis_
｜ 影响力: 18.6万粉丝 · 互动分 285.0
💡 核心观点: 引用：芯片出口正受审查，向中国出售数据尚未被纳入。有公司向中国卖出大量数据，其流程跑在前沿模型上，被作者视为一种蒸馏。
可信度: 55/100
事实核查:
- 原文支持: Chips are under scrutiny. Data sales to China aren't, yet.（原文直接陈述当前审查集中在芯片，数据销售暂未受限。）
- 无法核实: Selling and renting chips to China will face a lot more scrutiny.（这是对后续监管走向的预测，原文未给出依据。）
原文内容:
Chips are under scrutiny. Data sales to China aren’t, yet. “A lot of the data companies sell tons of data to China. It’s a really funny way to distill. Their pipelines run on frontier models, so if you’re building environments with those models and selling them to China, in some way you’re distilling the info out of them.” “Selling and renting chips to China will face a lot more scrutiny. Selling data hasn’t had any yet, and it will probably fall under that too.”
⏰ 21:18 | ❤️ 195 点赞 | 📝 89词 | 入选: 高赞 195 | 查看原文 →
AI 视频
MiniMax H3 以限时免费开放无限使用，主打局部编辑角色、动作与声音。
Hasan Toor
｜ @hasantoxr
｜ 影响力: 43.6万粉丝 · 互动分 190.8
Hasan Toor
｜ @hasantoxr
｜ 影响力: 43.6万粉丝 · 互动分 190.8
💡 核心观点: MiniMax H3 将于 11 月 19 日美东时间 8 点在 Envato 开放 24 小时免费无限使用，可在不改动场景其余部分的情况下编辑角色、动作与声音。
可信度: 68/100
事实核查:
- 原文支持: Envato is opening MiniMax H3 to everyone for free（推文与引用推文给出同一时间与免费范围。）
- 无法核实: H3 can edit characters, motion and sound without touching the rest of the scene（产品能力描述来自推广方，原文无演示或测试细节。）
原文内容:
If you had unlimited AI video for 24 hours, what would you make first? You'll get the chance on Nov 19. Envato is opening MiniMax H3 to everyone for free, starting at 8AM EST. H3 can edit characters, motion and sound without touching the rest of the scene, which is exactly what makes a free trial like this useful.
引用 @envato: Coming soon on Envato. Free unlimited @MiniMax_AI H3 for 24 hours. For everyone.*
Starting from 19 Nov at 8AM EST.
Create a free account to get access. Get ready to create. https://t.co/Yj5rlAQv3H
⏰ 13:08 | ❤️ 159 点赞 | 📝 55词 | 入选: 高赞 159 / 引用讨论 | 查看原文 →
行业动向
数学领域的冲击被视为即将传导到 AI 研发本身，安全评估与厂商对话同步推进。
Dwarkesh Patel
｜ @dwarkesh_sp
｜ 影响力: 28.7万粉丝 · 互动分 3765.0
Dwarkesh Patel
｜ @dwarkesh_sp
｜ 影响力: 28.7万粉丝 · 互动分 3765.0
💡 核心观点: The avalanche that hit math is coming for AI development.（规则摘要）
可信度: 50/100
原文内容:
The avalanche that hit math is coming for AI development. https://t.co/z5RsJgcWWc https://t.co/Ud6MDxiADh
⏰ 18:37 | ❤️ 2904 点赞 | 📝 16词 | 入选: 高赞 2904 | 查看原文 →
Anthropic
｜ @AnthropicAI
｜ 影响力: 190.7万粉丝 · 互动分 2947.0
Anthropic
｜ @AnthropicAI
｜ 影响力: 190.7万粉丝 · 互动分 2947.0
💡 核心观点: Streaming live at 12 PM ET: A fireside chat with JPMorgan CEO Jamie Dimon and @DarioAmodei.（规则摘要）
可信度: 50/100
原文内容:
Streaming live at 12 PM ET: A fireside chat with JPMorgan CEO Jamie Dimon and @DarioAmodei. https://t.co/9VFV9qgGjt Legal disclaimer: Anthropic has filed a registration statement (including a prospectus) with the SEC for the offering to which this communication relates. Before you invest, you should read the prospectus in that registration statement (available at: https://t.co/5sxoLV40ke) and other documents Anthropic has filed with the SEC for more complete information about Anthropic and this offering. You may get these documents for free by visiting EDGAR on the SEC web site at https://t.co/rBijFQGHi6. Alternatively, copies of the preliminary prospectus may be obtained from: Goldman Sachs & Co. LLC, Attention: Prospectus Department, 200 West Street, New York, NY 10282, by telephone at (866) 471-2526, or by email at [email protected]; or Morgan Stanley & Co. LLC, Attention: Prospectus Department, 180 Varick Street, 2nd Floor, New York, NY 10014. The securities may not be sold, nor may offers to buy be accepted, prior to the time the registration statement becomes effective. This post is not an offer to sell, or a solicitation of an offer to buy, the securities in any jurisdiction where the offer or sale is not permitted.
⏰ 16:00 | ❤️ 2035 点赞 | 📝 194词 | 入选: 高赞 2035 / 长文 | 查看原文 →
Anthropic
｜ @AnthropicAI
｜ 影响力: 190.7万粉丝 · 互动分 2645.0
Anthropic
｜ @AnthropicAI
｜ 影响力: 190.7万粉丝 · 互动分 2645.0
💡 核心观点: Safety evaluations are critical for understanding whether a model is ready to deploy, which safeguards to build for it, （规则摘要）
可信度: 50/100
原文内容:
Safety evaluations are critical for understanding whether a model is ready to deploy, which safeguards to build for it, and where to focus training and research. On our Engineering Blog, we offer 10 recommendations for how to design, run and report them. https://t.co/MtSQmqqi2D
⏰ 19:00 | ❤️ 1676 点赞 | 📝 44词 | 入选: 高赞 1676 | 查看原文 →
Chubby♨️
｜ @kimmonismus
｜ 影响力: 15.9万粉丝 · 互动分 2559.0
Chubby♨️
｜ @kimmonismus
｜ 影响力: 15.9万粉丝 · 互动分 2559.0
💡 核心观点: OpenAI has yet to release GPT-6.2.（规则摘要）
可信度: 50/100
原文内容:
OpenAI has yet to release GPT-6.2. There can only be three reasons: 1) They were surprised by how capable Fable 5.2 is and had to make GPT-6.2 better. 2) They found security vulnerabilities or other issues that need to be fixed. 3) Anthropic is going public next week, and OpenAI knows GPT-6.2 surpasses both Fable 5.2 and Opus 5.5. They’re trying to steal the spotlight ahead of their own planned IPO in 2027. I’m torn between 1 and 3, but I honestly don’t know. My gut tells me OpenAI was surprised by how good Fable 5.2 is. Either way, I expect GPT-6.2 to be released before November 19.
⏰ 16:35 | ❤️ 2014 点赞 | 📝 100词 | 入选: 高赞 2014 | 查看原文 →
🔗 引用关系
- @rasbt 引用 @rasbt：作者观察到各大厂商在一个多月内密集推出各自的 decision model，并引用 Vercel 报告称 26% 的请求已通过 decision model 路由。（原文：Took a while, but here it is, a mega write-up on language models for text classification and, of course, Jev.
It's basi…）
- @Hesamation 引用 @elonmusk（原文：@yunta_tsai A petition is underway to reinstate Jensen and I’m signing it too）
- @svpino 引用 @jakubjurovych：Deepnote 发布 Agent Initiative Benchmark，在模拟公司的 Slack 中测试 12 个前沿模型；代理可访问 11 项服务的记录，面对 50 种情境，须在 3 分钟内判断是否介入、如何帮忙、何时保持沉默。（原文：What’s the best model for proactive agents?
Today, we’re introducing the Agent Initiative Benchmark. We tested 12 front…）
- @DataChaz 引用 @xadisingh：Reason（reasonmachines，YC F26）发布编码代理 harness，宣称成本最多降低 20 倍，并在 Terminal-Bench 与 DeepSWE 上胜过顶级 harness；提供 Slack 集成、API 与 macOS 应用，可与 Claude Code、Codex、Cursor 并行运行。（原文：Frontier coding performance at up to 20x lower cost.
Introducing the world’s most cost-efficient harness: Reason
Stop …）
- @hasantoxr 引用 @envato：MiniMax H3 将于 11 月 19 日美东时间 8 点在 Envato 开放 24 小时免费无限使用，可在不改动场景其余部分的情况下编辑角色、动作与声音。（原文：Coming soon on Envato. Free unlimited @MiniMax_AI H3 for 24 hours. For everyone.*
Starting from 19 Nov at 8AM EST.
Cre…）
🧩 跨推文归纳
共识
- 评测侧多份独立结果指向同一趋势：Coding Agent Index 与图像编辑榜单同步更新，细分榜单上开源权重模型的位置正在抬升。
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
文章版权归作者所有，未经允许请勿转载。
相关文章
暂无评论...