888 X 動態摘要｜07/10

生成時間: 2026-07-10 05:03:09

總結

OpenAI與xAI近期動態顯示，兩大巨頭皆在AI模型性能、訓練效率、成本效益與產品商業化方面取得顯著進展，特別聚焦於企業級應用和生產力工具市場。

今日重點

1. OpenAI新模型高效低成本

• 人物: Greg Brockman (OpenAI 聯創)
• 時間: 3h
• 熱度: 👀 61,938
• 觀察: OpenAI推出GPT-5.6及Sol、Terra/Luna等新模型，在編碼、知識工作、資安與科學等多領域表現出色，且具備更優異的成本效益。
• 意義: 此進展顯示AI模型持續朝高效能低運營成本發展，有助於降低企業使用門檻，加速AI技術的普及與規模化部署。

2. xAI訓練效率獲重大突破

• 人物: Elon Musk (xAI 創始人)
• 時間: 21h
• 熱度: 👀 968,632
• 觀察: xAI透過C語言重寫核心訓練架構，使模型訓練效率大幅提升33%，特別在RL訓練階段可縮短兩週以上，且模型越大時間節省越顯著。
• 意義: 訓練效率是AI軍備競賽關鍵，此技術突破能讓xAI更快推出更大、更先進的模型，建立競爭優勢，並可能影響未來AI模型的迭代速度和市場格局。

3. Grok 4.5性能成本具領先

• 人物: Elon Musk (xAI 創始人)
• 時間: 3h
• 熱度: 👀 618,659
• 觀察: Grok 4.5在AA-Briefcase基準測試中，展現頂尖的代理式知識工作與編碼能力，同時實現極低的任務成本（$1.12）和更快的完成時間。
• 意義: 該模型在性能、成本與速度上的平衡，對於需高效能且預算受限的企業應用場景具強大吸引力，預示著高性價比AI解決方案的商業化潛力。

4. OpenAI積極回應成本顧慮

• 人物: Sam Altman (OpenAI CEO)
• 時間: 3h
• 熱度: 👀 339,257
• 觀察: OpenAI正透過GPT-5.6 Sol、Terra及Luna等新模型，顯著降低單位任務成本，積極解決企業用戶對AI應用成本的普遍擔憂。
• 意義: 成本是AI大規模企業採用的主要障礙，此進展直接回應市場需求，有助於擴大OpenAI企業客戶基礎和加速其商業化進程。

5. ChatGPT Work工作代理發布

• 人物: Greg Brockman (OpenAI 聯創)
• 時間: 3h
• 熱度: 👀 117,342
• 觀察: OpenAI發布結合ChatGPT與Codex的「ChatGPT Work」工作代理，旨在處理複雜任務，並支援移動、網頁與桌面等多平台使用。
• 意義: 這是OpenAI將通用AI與編程AI整合、推出企業級工作代理的具體商業化行動，預計將提升生產力工具市場競爭，並擴展AI應用場景。

6. OpenAI多產品線佈局

• 人物: Sam Altman (OpenAI CEO)
• 時間: 3h
• 熱度: 👀 600,259
• 觀察: OpenAI在GPT-5.6發布直播中，同步推出ChatGPT Work、全新ChatGPT桌面應用程式及託管網站服務等多項產品更新。
• 意義: 這些產品更新顯示OpenAI正積極擴展其生態系統和商業化佈局，透過多平台應用和服務拓展市場，以提升用戶粘性及競爭力。

原始動態

重點 | Greg Brockman

• 時間: 3h
• 熱度: 👀 61,938
• 原文: GPT-5.6 is here. Sol is an incredible model, and Terra/Luna provide great performance at lower price. Great at coding, knowledge work, cybersecurity, and science with fewer tokens and at lower cost. openai.com

重點 | Elon Musk

• 時間: 21h
• 熱度: 👀 968,632
• 原文: "Grok 4.5 is looking like a success with help from Cursor data but underneath the surface we expect future Grok/Cursor model training is likely to speed up in the coming months. We've spent several months getting up to speed on the SpaceXAI business, especially the tech underneath it. The C rewrite was under appreciated by the investor community so we dug in to quantify its impact. including building a physics first model that functions as a stopwatch for the SpaceXAI model factory. Bottomline: C-rewrite gets SpaceXAI faster model cycles, leveraging 33% more tokens/second/GPU against SOTA competition resulting in the potential to shipping new models every ~3.5 weeks. two core learnings from this modeling exercise: 1: the training cycle speed up is primarily coming from RL (not pretraining) where the increased tokens/second/GPU advantage can shave up 2+ weeks off full model training cycle. 2: the rewrite itself should compound the time savings as model sizes grow. at ~2T shaving off 2-3 weeks, ~8 weeks at 6T, and 15 weeks at 10T. Note: a 20T parameter model likely runs into a data bottleneck prior to a training speed bottleneck but the directional advantage stands. Also we assume tokens/second/gpu advantage will melt over time as competitors try to match it. when you do the math, in true SpaceX and Elon fashion, it looks like they are attempting to build a SOTA model factory that can pump out bigger models faster than anyone else. Full analysis here for the public:"

重點 | Elon Musk

• 時間: 3h
• 熱度: 👀 618,659
• 原文: Grok 4.5 is the top non-Anthropic model on AA-Briefcase, combining frontier agentic knowledge work capabilities with leading cost and time-efficiency Yesterday "released Grok 4.5, a new frontier-level model with strengths in agentic coding and knowledge work. On AA-Briefcase, Grok 4.5 scores 1328, a +578 improvement over Grok 4.3 and the highest score of any non-Anthropic model (note that GPT-5.6 not released yet). It achieves this while sitting on the cost and time efficiency frontier, averaging $1.12 per task, 86% lower than Claude Opus 4.8 (max), and 12.4 minutes per task, around half the time of Opus 4.8 (max). AA-Briefcase is our new proprietary benchmark for agentic knowledge work, testing models on a fully private dataset of realistic tasks across thousands of complex input files. Tasks require deliverables like spreadsheets, presentations, and UI mock-ups, with performance combined into a single AA-Briefcase Elo across correctness, analytical quality, and presentation quality. Key results for Grok 4.5 with high reasoning on AA-Briefcase: ➤ Frontier agentic knowledge work capabilities: Grok 4.5 achieves an AA-Briefcase Elo of 1328, the highest score of any non-Anthropic model, behind only Claude Fable 5 (1390), Claude Sonnet 5 (max, 1390), and Claude Opus 4.8 (max, 1354). Across the three AA-Briefcase scoring axes, Grok 4.5 is strongest on objective rubric criteria and analytical quality, with comparatively weaker presentation quality. It achieves the second-highest overall rubric pass rate (40.7%), behind Claude Fable 5 (56%) and Claude Sonnet 5 (42.3%) ➤ Leading cost efficiency: Grok 4.5 averages a cost of $1.12 per AA-Briefcase task, placing it on the cost-performance Pareto frontier. This is much most cost effective than peer models such as Claude Opus 4.8 (max, $8.26) and GLM 5.2 (max, $1.71) ➤ Faster task completion: Grok 4.5 averages 12.4 minutes per AA-Briefcase task, also placing it on the speed-performance frontier. It is much faster than Claude Opus 4.8 (max, 23.9 min) and Claude Sonnet 5 (max, 36.9 min), primarily due to lower turn use. Grok 4.5 averages just 23 turns per task, ~40% of GLM 5.2 (max, 56) and ~13% of Claude Sonnet 5 (max, 183) Congratulations to" "," ", and" on the impressive release!

重點 | Sam Altman

• 時間: 3h
• 熱度: 👀 339,257
• 原文: we have heard enterprises on their concerns about AI costs, and 5.6 sol is a huge step forward for dollars-per-task, as are terra and luna

其他 | Sam Altman

• 時間: 1h
• 熱度: 👀 165,473
• 原文: check this out! you can get some amazing things done. codex is the core of our new work product and what makes it so good. codex is not going anywhere.

重點 | Greg Brockman

• 時間: 3h
• 熱度: 👀 117,342
• 原文: "We’ve brought together ChatGPT and Codex, in the form of ChatGPT Work: an agent for your most ambitious work. Use it from mobile or web, in addition to desktop — no need to leave your laptop cracked open!"

重點 | Sam Altman

• 時間: 3h
• 熱度: 👀 600,259
• 原文: 5.6 livestream going now. in addition to the model, 3 major product things. 1. ChatGPT Work--really big deal! 2. new ChatGPT desktop app 3. hosted sites

其他 | Sam Altman

• 時間: 16h
• 熱度: 👀 581,091
• 原文: tbh i dont think sol gets that many dates either

其他 | Ray Dalio

• 時間: 2h
• 熱度: 👀 40,165
• 原文: I often observe people making decisions if their odds of being right are greater than 50 percent. What they fail to see is how much better off they'd be if they raised their chances even more (you can almost always improve your odds of being right by doing things that will give you more information). The expected value gain from raising the probability of being right from 51 percent to 85 percent (i.e., by 34 percentage points) is seventeen times more than raising the odds of being right from 49 percent (which is probably wrong) to 51 percent (which is only a little more likely to be right). Think of the probability as a measure of how often you're likely to be wrong. Raising the probability of being right by 34 percentage points means that a third of your bets will switch from losses to wins. That's why it pays to stress-test your thinking, even when you're pretty sure you're right.

其他 | Balaji Srinivasan

• 時間: 4h
• 熱度: 👀 223,777
• 原文: We’re building Silicon Valley outside Silicon Valley.

其他 | Eric Topol

• 時間: 7h
• 熱度: 👀 34,765
• 原文: The science of human aging is flourishing, perhaps best exemplified by remarkable advances in organ and cellular clocks, tracked from proteins in the blood. These clocks tell us about the pace of aging within an individual and are linked to healthspan, longevity, and diseases. and I reviewed the field of biological clocks, published today free access

其他 | Sam Altman

• 時間: 16h
• 熱度: 👀 531,182
• 原文: what a good video

其他 | Thomas Wolf

• 時間: Jul 6
• 熱度: 👀 47,136
• 原文: Today we’re announcing our “ThinkingCap” efficient model series with a 2× thinking token reduction on average in Qwen 3.6 27B, with up to 10x faster generation on individual examples.

其他 | Andrew Ng

• 時間: 5h
• 熱度: 👀 28,577
• 原文: "We stand at a critical crossroads in the debate over AI governance in the United States, and it feels like we are inching closer to a very serious battle over whether or not open source models will even be allowed in an environment where a new de facto licensing regime has been taking shape. Lacking formal congressional statutory frameworks or clear administration rules (like the diffusion rule revision), we appear to be left with a sporadic, arbitrary, non-transparent process for model review. The fiction of “voluntary” agreements hangs over this debate, and some large model developers are already showing an incredible willingness to bend over backwards to accommodate national security-related officials / orders that the rest of us are not privy to. It's a very opaque process. And those model developers are expected to play ball with those officials, or else their models get pulled from the market or held up for long periods. Or they will lose any government procurement contracts they have. There is nothing “voluntary” about it when that Sword of Damocles hangs in the room. As this mess worsens, at some point the question of how to handle open source models will come into sharper focus because it will have to. I've even heard some rumors lately that something may be coming from the admin on this front to address this. Needless to say, if this informal new AI model review regime expands and takes on more pre-vetting characteristics / requirements, it is hard to see how open source players could comply with such quasi-licensing of AI models. Specifically, if this ambiguous new regime is accompanied by a general presumption of ‘restrict-until-permitted,’ then that would spell doom for open source. That is a very dark path for our country. Worse yet, of course, would be a move by national security officials to more directly restrict open source models and capabilities. If that happens, then we would be right back in the thick of a Clipper Chip-like battle along the lines of what we saw in the late 1990s. That is a much darker path for America. Meanwhile, open source developers have no “golden shares” or other goodies to offer the government to make their problems go away. Let’s be clear: If our government takes the dark path, it will become the single most important battle over computational freedom of modern times. It is time for people to make a stand in defense of open source before it is too late.

其他 | Balaji Srinivasan

• 時間: 3h
• 熱度: 👀 23,173
• 原文: James is from Great Britain. We have many Brits at Network School, and indeed many from the global Anglosphere (including the UK, Canada, Australia, and New Zealand). All working together to build positive-sum, in-person communities.

其他 | Balaji Srinivasan

• 時間: 4h
• 熱度: 👀 21,353
• 原文: The purpose of Network School is to demonstrate that anyone can peacefully start new communities from the Internet. Our vision of the good is a society that's tolerant and meritocratic, affordable and ambitious, high-tech yet extremely offline.

其他 | Huanusa

• 時間: 21h
• 熱度: 👀 10,197
• 原文: 养生这件事，第一步不是买补品。 而是先管住嘴。 很多食物不是不能吃，而是别长期、频繁、大量吃。 下面这 40 类，建议少碰。 1. 猪头肉 高脂肪、高胆固醇，偶尔吃可以，别当日常下酒菜。 2. 炒粉 淀粉裹油，高温爆炒后热量很高，吃一盘很容易超量。 3. 油炸花生米 看着小，油脂不低。越香越容易停不下来。 4. 腊肉腊肠 高盐腌制食品，偶尔解馋即可，别长期当主菜。 5. 动物内脏 胆固醇和嘌呤含量相对高，吃的时候要控制频率和分量。 6. 奶油蛋糕 糖分、脂肪都高，生日吃一块可以，别经常当下午茶。 7. 果脯蜜饯 水果经过糖渍后，很多时候已经变成高糖零食。 8. 速溶咖啡 尤其是三合一速溶咖啡，常见问题是糖和植脂末偏多。 9. 膨化食品 高油、高盐、低营养，越吃越想吃，最容易不知不觉超量。 10. 珍珠奶茶 奶精、糖、木薯淀粉叠加，一杯下去热量不低。 11. 散装腌菜 来源和保存条件不清楚，尽量少买。自家腌制也要注意时间和卫生。 12. 加工肉丸 有些肉丸淀粉、增稠剂、调味剂不少，买之前先看配料表。 13. 啤酒烧烤 炭烤、高油、高盐，再加酒精，偶尔可以，别变成固定宵夜。 14. 功能饮料 咖啡因和糖分可能偏高，非必要不要长期依赖。 15. 果粒酸奶 很多果粒酸奶含糖量并不低，真正想补充乳制品，优先选原味。 16. 精制面包 白面粉、糖、油脂偏多，早餐长期吃不如换全麦、鸡蛋、牛奶。 17. 火锅丸子 加工度高，久煮还会吸油吸盐，少吃为妙。 18. 水果罐头 糖水浸泡后，热量上升，营养价值不如新鲜水果。 19. 辣条 重盐、重油、重调味，偶尔回忆童年可以，别常吃。 20. 勾芡汤汁 浓稠不等于营养，很多汤汁是淀粉、油脂、盐分的组合。 21. 含糖酸奶 酸奶本身已有乳糖，再额外加糖，容易变成甜品。 22. 话梅瓜子 盐分高，越吃越口干，也容易摄入过量钠。 23. 自助餐 最大问题不是食材，而是容易吃过量。七分饱就停。 24. 浓白高汤 汤越白，不一定越补。很多时候是脂肪乳化后的效果。 25. 果蔬脆片 看着像蔬菜水果，本质可能是油炸脱水零食。 26. 含乳饮料 别当牛奶喝。很多只是水、糖、添加剂和少量乳成分。 27. 蛋黄派类零食 代可可脂、人造奶油、糖分都不少，应急可以，别囤着天天吃。 28. 油炸春卷 外皮和馅料都容易吸油，年节吃一点即可。 29. 水果月饼 别被“水果”两个字迷惑，很多馅料仍是糖油混合物。 30. 散装坚果 坚果怕氧化，散装放太久容易变味，优先选小包装。 31. 方便面汤 面可以偶尔吃，汤别喝到底，钠含量通常不低。 32. 西式快餐 炸鸡、汉堡、薯条、可乐组合，长期高频吃容易热量超标。 33. 红糖馒头 别把它当健康食品。精制面粉还是主体，升糖也不低。 34. 果蔬汁饮料 喝果汁不如直接吃水果。榨汁后膳食纤维减少，糖吸收更快。 35. 冰镇啤酒 冷饮加酒精，对胃肠刺激较大，尿酸高的人更要谨慎。 36. 糯米制品 粽子、汤圆、年糕消化慢，胃容易胀的人要少吃。 37. 散装香油 来源不明的低价香油要谨慎，尽量买正规小瓶装。 38. 过夜饭菜 剩菜要及时冷藏、充分加热。绿叶菜和海鲜更不建议反复隔夜。 39. 含糖茶饮 瓶装茶、调味茶饮很多糖分不低，不如自己泡茶。 40. 网红零食 包装越花，越要看配料表。别被“低卡”“健康”“无负担”几个字带节奏。 真正的养生，不是完全不吃。 而是知道什么东西不能天天吃。 少油、少糖、少盐、少加工。 多吃原型食物， 少吃配料表很长的东西。 嘴管住一半， 身体负担就少一半，尤其国内的朋友食品安全相对恶劣的情况下，真得自己多一分警惕才好。

其他 | Eric Topol

• 時間: 2h
• 熱度: 👀 5,368
• 原文: Called Biomni, its performance surpassed LLMs it was assessed against and exhibited capability for diverse tasks that include drug repurposing, rare disease diagnosis, single-cell annotation, causal gene identification, molecular cloning, protein stability, design and orchestrating wet-lab experiments, and more

其他 | Ray Dalio

• 時間: Jul 8
• 熱度: 👀 69,808
• 原文: Make sure it's clear how much weight each person's vote has so that if a decision must be made when there is still disagreement, there is no doubt how to resolve it.

其他 | Huanusa

• 時間: 6h
• 熱度: 👀 8,251
• 原文: 《大明王朝1566》最狠的地方在于： 它把很多人情世故、权力逻辑、职场规律，都讲透了。 以前看到一句话： 小事上从不考虑你的人，大事上也不会突然选择你。 如果某天突然想起你了， 大概率不是要给你好处， 而是要你去担风险。 放到《大明王朝1566》里看，非常明显。 海瑞之前一直在南平当教谕， 没什么人真正把他放在核心位置。 可一到“改稻为桑”这种棘手事情， 突然就被推上来了。 海瑞母亲看得很清楚： 那么多大官不去争， 为什么偏偏让一个小小县令去争？ 这不是重用。 这是让他去顶压力。 高翰文也是一样。 平时在翰林院过清苦日子， 没多少人真正惦记他。 可“改稻为桑”一来， 突然被提拔， 突然被委以重任。 看似是机会， 其实是把他放进了风暴中心。 再看马宁远。 平时杨金水他们根本不会想起这个人。 可一到毁堤淹田这种大事， 突然请他吃饭， 突然安排庆功宴， 突然让他办事。 这哪里是看重？ 这就是提前找好替罪羊。 事成了， 功劳未必有他。 事败了， 锅大概率扣他头上。 所以很多时候， 一个人突然被“重用”，未必是好事。 关键要看： 这件事有没有真正的权力支持？ 有没有资源？ 有没有退路？ 出了问题谁负责？ 如果只给任务， 不给权力； 只给风险， 不给保护； 只让你冲锋， 不让你参与分配， 那就不是提拔， 而是消耗。 真正成熟的人， 不会因为别人突然说“这事交给你最合适”，就立刻感动。 而是先想清楚： 为什么是我？ 为什么现在是我？ 如果这么重要， 以前为什么不是我？ 《大明王朝1566》之所以耐看， 就是因为它讲的不是古代故事， 而是人性和权力的底层逻辑。 很多局， 换个时代， 照样成立。

其他 | Huanusa

• 時間: 5h
• 熱度: 👀 2,464
• 原文: “美团自动删除用户图片” 其中这事讨论最热的是3月19日，山东，一网友称，美团app侵犯公民隐私，私自删除了自己手机相册中的2千多张照片。 评论区也有部分网友反映相同的情况 实际国内很多APP都有这个问题尤其是最近又被剧烈讨论的微信！

其他 | Huanusa

• 時間: 17h
• 熱度: 👀 1,761
• 原文: 布兰登综合胜率35%，大家觉得这数据靠谱吗？ tianjifiles.com

其他 | Greg Brockman

• 時間: Jul 8
• 熱度: 👀 6,600,786
• 原文: Introducing GPT-Live, a new generation of voice models for natural human-AI interaction. Rolling out in ChatGPT starting today. You’ll want to turn the sound on for this one.

其他 | Greg Brockman

• 時間: 4h
• 熱度: 👀 1,337,322
• 原文: Today. 10am PT.

其他 | Peter Steinberger

• 時間: 7h
• 熱度: 👀 119,378
• 原文: Big day for Ollama! When we started, open models and the open source AI ecosystem were in their early days with few believers. Our belief in open source has never wavered. With today's fundraising announcement and our 9M+ active builders, we’re ready to scale open models into AI that you can own. All aboard open models! 🧵

其他 | Nassim Taleb

• 時間: 9h
• 熱度: 👀 133,461
• 原文: Interesting observation.

其他 | Peter Steinberger

• 時間: 9h
• 熱度: 👀 37,188
• 原文: Ya know the little one-liners when OpenClaw starts up? Time for some new ones.

其他 | Nassim Taleb

• 時間: Jul 8
• 熱度: 👀 89,512
• 原文: On July 8, 1972 the Israeli terrorist state placed a bomb under the car of Palestinian writer/journalist/artist Ghassan Kanafani. They killed him along with his 14-year old niece. You think decent human beings forget and forgive those terrorist crimes by Israel?

其他 | Greg Brockman

• 時間: 3h
• 熱度: 👀 56,297
• 原文: happening now!

其他 | Huanusa

• 時間: 10h
• 熱度: 👀 23,805
• 原文: 这才是中国梦 Made with AI

其他 | Eric Topol

• 時間: 3h
• 熱度: 👀 9,420
• 原文: The generations of biological clocks to track human aging

其他 | Sean Kelly

• 時間: 17h
• 熱度: 👀 201,105
• 原文: One of physicists' most popular recent idea, that entanglement is really evidence for wormholes, just died

其他 | Eric Topol

• 時間: Jul 8
• 熱度: 👀 35,638
• 原文: Today's sleep quality tracking relies on movement algorithms. Published today is a wearable patch that tracks the flow of glymphatics via brain water dynamics