The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Claude models is not affected. We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible. Read our full statement: anthropic.com/news/fable-myt…

GLM-5.2 is Fully Open, Frontier Intelligence Belongs to Everyone Today, the sudden restriction of certain frontier models is deeply regrettable. At a time when access to frontier models is abruptly cut off for non-technical reasons, we are even more convinced of one thing: science should be global. The path to AGI (Artificial General Intelligence) must never be enclosed by high walls. We have always believed that AGI should be the cornerstone for all of humanity to collaboratively explore the boundaries of intelligence and solve complex challenges, rather than a privilege monopolized by a few rules and subject to revocation at any moment. In the face of external blockades and restrictions, our attitude is one of radical openness. Frontier intelligence must remain open-source, accessible, and buildable, serving every dedicated developer. GLM-5.2 is Zhipu's most capable open-source model to date. It not only supports a truly usable 1M context window but also maintains a continuous lead in the independent completion of long-horizon tasks, providing solid foundational support for building complex agent applications. It also continues to be our main engine for creating the strongest domestic coding model. Tonight at 5:21—at this special moment—GLM-5.2 will officially be available to all GLM Coding Plan users (including Lite / Pro / Max). The API will also go live next week. A step closer to frontier intelligence for everyone. The future of AI is open, and it is for the people. ModelKey: GLM-5.2

If you've adopted AI at your company but haven't seen any tangible results, read this 1990 article: "The Dynamo and the Computer" by Paul David. When electricity first arrived, factories that "adopted" it barely got faster. They just swapped the steam engine for an electric one and ran everything else exactly as before: same machine layout, same workflow, same management. Electricity in, no real gains out. The most common mistake with any new technology is to drop it into the old organization and then declare the transformation done. The real leap came decades later, when each machine got its own small motor. Suddenly machines no longer had to be lined up around one central drive shaft. They could be rearranged around the actual flow of work. The productivity gains didn't come from electricity. They came from REDESIGNING THE ENTIRE FACTORY around it. AI is the same. Bolting it onto your existing process gets you a faster steam engine. The payoff comes when you redesign the work itself. (link to paper in comments)

BREAKING NEWS: Anthropic's latest model will NOT help you if it thinks your ML research/ML engineering is interesting, and/or will secretly degrade its IQ so that the average engineer won't notice. We are already seeing Anthropic's latest model's moderation filters our GPU inference research and programming 😭

Tesla AI chip design engineering reviews are so great! Team is awesome. Our AI6 chip might set a record for most amount of usable intelligence from a wafer when factoring in yield.

有个事挺有意思的。 DeepSeek V4 的技术报告，对所有主流大模型做了一轮横评，结论是——Gemini 3.1 Pro 的世界知识是所有模型里最强的。 不是 GPT，不是 Claude，是 Gemini。 但大家用 Gemini 的感受普遍是：这玩意好用吗？ 问题不在模型本身，在于它极度懒得动。 你要问它最新的新闻，它有搜索工具，但就是不主动用。很多时候你得明确说你去搜一下，它才搜。就像一个博览群书的人，你问他最近发生了什么，他耸耸肩：我没看今天的报纸。 一个世界知识最强的模型，工具懒得调——这才是 Gemini 用起来别扭的真正原因。

A\ 蒸馏国产是啥新鲜事吗😅 从 4.5 开始，没有 system prompt 的 Claude API 都在说自己是 Deepseek 啊😅 甚至你现在去 Vertex 用中文问 sonnet-4.6 他也说自己是 Deepseek v3 这是正版的标识 AI 这一行就是蒸蒸日上 唯有某个婊子天天打舆论战立牌坊

Most of AMD and NVIDIA’s best 10x engineers are in Shanghai. AMD’s MoRI collective team, AMD’s disaggregated applications engineering team, and other AMD teams that understand how to do first-principles-based engineering are all mostly based in Shanghai.

If time travel becomes possible, what year would you visit first and what tech would you take with you? ✍️

我的思维还是有问题，不敢给 AI 提需求，也许是惯性思维影响，因为以前一提开发需求，就是要对应的资源，但是自从用 codex 后，大的需求也是非常谨慎，其实这是错的，应该大胆的给 codex 提需求。

Meta was easily the most toxic company I've worked for. There's a reason the Chinese call it "Squid Game". Others refer to it as "Hunger Games" or "Lord of the Flies". I think they're all accurate. The company culture is basically every man/woman for themselves. The performance review process (PSC) not only doesn't incentivize helping others, if anything it actually discourages it since everyone is stack ranked against each other. Imagine working on a team where every 6 months, one of you is going to get axed. Of course it's going to become toxic. "Bottoms up" culture is a complete farce - it's just a way for leadership to offload accountability. The Tech Leads (TLs) have all the power - owning the relationships and tribal knowledge to gatekeep projects to their buddies. Managers are "people managers" with limited technical understanding, who basically aggregate TL feedback and create performance review packets to calibrate with other managers and IC7 . The takeaway is that your destiny is in the hands of the TLs, and TLs unlike managers have no responsibility for your career. There are no repercussions for unethical behavior. I've seen managers and TLs throw others under the bus and get away with it. The only mission bonding the company together is individual self-preservation. Save your own ass to survive for another stock vesting, and throw someone else under the bus if you need to. That's why layoffs rarely impact directors/VPs or tenured IC7 despite the fact that they're paid by far the most. Even this recent mass layoff that was supposed to "flatten" managers layers barely affected directors/VPs/IC7 , and fell predominantly on M1s - the lowest rung of the management chain. The culture is extremely performative and focused on box ticking and optics. Everything is about PSC (the performance review system) and perception. This means tons of meetings, useless AI slop posts, and top-down initiatives that don't benefit anyone but maybe help tick off the impact box of some go-getter at the top. Impact is not enough - it has to have sufficient complexity. So complexity is added for complexity's sake. The org I was in (Facebook ads) is 90% Chinese, and the entire leadership chain up to the VP level is Chinese. Mandarin is the primary language at the office, except in official meetings with non-speakers. Chinese work culture is very different from American work culture, with 996 (9am-9pm, 6 days/week), top-down nature, emphasis on saving face (eg. don't question your superiors), and toxicity being quite common. Naturally when an org is completely dominated by a single ethnicity that's notorious for not integrating, elements from their work culture seep in. Of the layoffs I witnessed in this org, 3/4 were not Chinese (just to be clear, most Chinese are very kind so don't take this as an attack. But it is a reality that I think most people outside this company are completely unaware of, and I question if leadership is even aware despite the fact that we're talking about the company HQ) I had the most toxic manager of my life here. I watched him deliberately set up a new hire to fail, driving them to needing to see a psychiatrist for anxiety depression, and getting them fired. Then he suddenly disappeared for 8 months, before leaving the company. I could go on and on, but this is already pretty long and I think you get the point. Yes there are a lot of great, kind people here. I managed to transfer out of my first team into a new team with a great manager where everyone was very smart, supportive, and hardworking. But the company has its Squid Game reputation for a reason. Company culture comes from the top. It seems leadership is either too removed to notice, or maybe don't really care anymore because I guess they already made their billions and us plebs are expendable these days.

很多人在说 AI 到瓶颈了。Scaling Law 不行了，模型能力到头了，接下来就是做应用了。 Karpathy 今天加入 Anthropic 做预训练，直接打了这个判断的脸。 他是全球最有资格判断"模型还能不能继续进化"的人之一。他的结论不是"差不多了"，而是"不回实验室我就要落伍了"。 你现在觉得 AI 做不到的事情，半年后可能就能做到了。你现在觉得不够好用的地方，一年后可能完全不是问题了。 千万不要用今天的能力边界，去画明天的天花板。模型还在快速进化，这条路远远没有走到头。

New blog post: The third wave of American philanthropy Hundreds of billions of dollars in new philanthropic capital will soon become liquid. The OpenAI Foundation holds 26% of OpenAI, worth about $220B at today’s valuation. Anthropic’s seven co-founders have pledged to give away 80% of their wealth and have instituted the most aggressive donor matching program for employees in tech history. How much does this all add up to? And how meaningful is that in the context of philanthropy today? I was doing some simple napkin math to wrap my head around the scale of what’s coming, and radicalized myself in the process. I had dramatically underappreciated the scale of the philanthropic capital that’s about to become available and the corresponding gap in talent and organizations that will be needed to make the most of it. This piece aims to directionally sketch the scale of what’s coming, the gap in operational capacity needed to absorb it, and what we can do to fill it. (Link to full post in reply)

The metrics on 3.5 Flash are major: 4x faster than other frontier models in output tokens per second Outperforms our previous 3.1 Pro model on nearly all benchmarks Shows massive improvement on coding and agentic benchmarks like Terminal-Bench 2.1 (76.2%), GDPval-AA (1656 Elo) and MCP Atlas (83.6%)