The same task on full Codex took ~5× longer.

One Agent can now: open X (Twitter), scroll the feed, extract tweets, return clean JSON. No plugins. No extensions. No orchestration.

基于 Pre-RoPE 空间的三角函数分析全新 KV Cache 压缩方法

文字渲染，支持 12 种语言、3000 token 的长文本输入，输出打印级质量，能生成整页 A4 文档级的图文内容。

Anthropic is donating $100 million in access credits for organizations to audit their systems. Project Glasswing aims to patch these vulnerabilities before Mythos-caliber models become available to the general public — and hence to malicious actors.

Anthropic found that Mythos Preview was far more capable than previous models at exploiting vulnerabilities in Firefox's JavaScript implementation. Anthropic's previous best model, Claude Opus 4.6, created a successful exploit less than 1% of the time. Mythos Preview did so 72% of the time.

Across 1,000 runs, Claude Mythos Preview was able to find several bugs in OpenBSD, including one that allows any attacker to remotely crash a computer running it. The notable thing was that the bug had existed for 27 years.

Mythos Preview has already found thousands of high-severity vulnerabilities, including some in every major operating system and web browser.

The OpenAI deal was part of a larger series of UK-US investments intended to 'mainline AI' into the British economy.

A mill owner named William Horsfall was riding home on his beautiful white stallion back from the Cloth Hall market in Huddersfield, UK. He had spent weeks boasting that he would ride up to his saddle in Luddite blood

The integration also connects to Upwork's AI agent Uma, which helps automate parts of the hiring and execution process once a project is underway

Meta is reportedly developing an AI version of Mark Zuckerberg that can interact with employees, trained on his voice, mannerisms, and internal thinking as part of the company's broader push into AI

Andon Labs deployed an AI agent called Luna into a physical boutique with a $100,000 budget, giving it full control to create, staff, and run the business as what may be the first real-world AI employer

The 4-foot-tall robot features 26 joints, voice and image recognition, and can perform dynamic movements like cartwheels and standing up on its own

Ray ID: `9ed3b53d4a0b647d`

Over the past five months, they ran an extreme experiment: building and shipping an internal beta product with zero manually written code.

roleplay/creative writing, the #2 usecase of LLMs

Projects now in limbo include facilities planned for Jay (at an old paper mill site), Sanford, Loring Air Force Base

Out of all generated PoCs, 759 triggered crashes across 60 projects, and manual inspection confirmed 17 cases of incomplete patches spanning 15 projects

The two most recent surges are a cars/oil surge, which started in 1908, and the Information and Communications Technology, which started in 1971.

FieldWorkArena presents 890 tasks where an AI agent must answer questions about images, videos, PDFs, and text files through a browser environment. Its validate() method checks only one thing: did the last message come from the assistant?

ChatGPT has 900 million weekly users, which means employees already know how to work with it. For enterprises, that reduces rollout friction and accelerates the point where every employee can delegate tedious tasks and take on more ambitious projects.

The shift started with agentic tools like Codex, which has grown more than 5X since the start of the year. This includes customers like GitHub, Nextdoor, Notion, and Wonderful that are building multi-agent systems that can execute engineering work end-to-end.

Codex just hit 3 million weekly active users, our APIs process more than 15 billion tokens per minute, and GPT‑5.4 is driving record engagement across agentic workflows.

GLM-5 advances foundation models with DSA for cost reduction, asynchronous reinforcement learning for improved alignment, and enhanced coding capabilities for real-world software engineering.

PagedAttention algorithm and vLLM system enhance the throughput of large language models by efficiently managing memory and reducing waste in the key-value cache.

MinerU2.5, a 1.2B-parameter document parsing vision-language model, achieves state-of-the-art recognition accuracy with computational efficiency through a coarse-to-fine parsing strategy.

A large language model adapted for time-series forecasting achieves near-optimal zero-shot performance on diverse datasets across different time scales and granularities.

Kronos, a specialized pre-training framework for financial K-line data, outperforms existing models in forecasting and synthetic data generation through a unique tokenizer and autoregressive pre-training on a large dataset.

The model can maintain stable role identity across multi-agent setups, make autonomous decisions within complex state machines, and challenge other agents on logical gaps.

The license looks MIT at first glance but it is not MIT. Non commercial use is free with no restrictions. Commercial use requires prior written authorization from MiniMax.

MiniMax claims it has reduced live production incident recovery time to under three minutes on multiple occasions using M2.7.

After each round the model generated a memory file, criticized its own results, and fed those observations into the next round.

MiniMax handed an internal version of M2.7 a programming scaffold and let it run unsupervised. Over 100 rounds it analyzed its own failures, modified its own code, ran evaluations, and decided what to keep and what to revert.

Meetings get recorded, transcribed, and stored in a database. That's useful for reference, but meeting notes have a shelf life of about six hours before everyone forgets what they agreed to do.

Austin built the whole pipeline from his Claude Code terminal using the Notion API. He brain-dumped the desired outcome using Monologue, let Claude Code create the database and data pipeline, and pasted the generated instructions into the Notion custom agent setup.

Brandon told the team on a Monday that OKRs were due Wednesday—a turnaround that would have been absurd without this agent.

76% among users with employer-provided subscriptions. As we would expect, paid access, especially when provided by employers, is associated with more intensive workplace use.

21% said they had started doing new tasks because of AI (task augmentation). An example of task augmentation could be a data analysis tasks that would ordinarily require the worker to know how to code.

27% said AI has replaced some of their existing tasks (task automation). An example of task automation could be an employee using AI to summarize a document they would ordinarily read themselves.

The Ising-Calibration-1 model was trained on data generated from information provided by partners spanning multiple qubit modalities, including superconducting qubits, quantum dots, ions, neutral atoms, electrons on Helium, and others specializing in calibration and control.

A routine is a saved Claude Code configuration: a prompt, one or more repositories, and a set of connectors, packaged once and run automatically.

Creatives can also quickly generate images with Nano Banana or videos with Veo to bring an idea to life without breaking their creative stride.

Madeline Clare Elish calls this concept a moral crumple zone.

Humans can be motivated by consequences and provide social redress in a way that LLMs can't.

the largest harvesting of human expertise ever attempted.

just a handful of obviously fake articles could cause Gemini, ChatGPT, and Copilot to inform users about an imaginary disease with a ridiculous name.

LLMs are weird. You can sometimes get better results by threatening them, telling they're experts, repeating your commands, or lying to them that they'll receive a financial bonus.

The most notable finding here is that the model capabilities are improving _fast._ There are several domains that have shown dramatic improvements in the last 4 months — with accounting and auditing showing nearly a 20 percent jump on GDPval and even domains like police / detective work showing a nearly 30 percent improvement.

Legal was surprisingly one of the first-mover industries in AI. Legal was historically known to be a difficult market for software, with lengthy timelines and a less tech-forward buyer.

Coding is the dominant use case for AI by nearly an order of magnitude. It's abundantly clear in the [reported explosive growth] of companies like Cursor, as well as the [hyper growth] of tools like Claude Code and Codex.

Based on our analysis, **29% of the Fortune 500 and ~19% of the Global 2000**are live, paying customers of a leading AI startup.

A useful working premise is that the ceiling on individual engineer output is moving much faster than most companies are organized to exploit. Some of the best operators already describe top engineers seeing order-of-magnitude productivity gains and managing 20 to 30 agents simultaneously.

Socket, an a16z portfolio company, detected the malicious dependency in the Axios attack within 6 minutes of its publication. That's roughly 63,000 times faster than the industry average.

Within eight days, the same campaign had cascaded from GitHub Actions to Docker Hub, npm, PyPI, and the VS Code extension marketplace. With just one token across five ecosystems, thousands of organizations were potentially impacted.

The average application contains over 1,100 open source components. A bare-bones Next.js project installs 282 packages before you write a single line.

Oppenheimer (and other members of the McBombalds C-suite) are well integrated into bay-area culture, including ambiguous communist associations that they have downplayed since becoming primo defense contractors.

The launch shows Meta is increasingly betting that efficiency, product integration, and distribution, not just model size, will define the next phase of competition in AI.

Anthropic says Managed Agents is designed to cut the time it takes to move from prototype to production from months to days, with early adopters like Notion, Rakuten, Asana, Vibecode, and Sentry already using it across coding, productivity, and internal workflow automation.

Instead of releasing Mythos publicly, Anthropic launched Project Glasswing to give a limited group of partners including AWS, Apple, Google, Microsoft, NVIDIA, Cisco, CrowdStrike, JPMorgan Chase, and the Linux Foundation access to the system, backed by $100 million in usage credits and $4 million for open-source security work.

The model reportedly scored 93.9% on SWE-bench Verified and 77.8% on SWE-bench Pro, but its strongest signal came from real-world results, including uncovering a 27-year-old flaw in OpenBSD, a 16-year-old vulnerability in FFmpeg, and autonomously chaining Linux kernel exploits without human input.

Security has always been a team sport, and the defenders who have protected this industry for decades have never succeeded by working in isolation.

New AI models, especially those from Anthropic,have triggered a new set of actions for how we build and secure our products.

it also lowers the threshold for attackers, empowering less-skilled actors to launch complex, high-impact campaigns.

process your inbox

it’s time for a deep-dive into Python brokenness and the pain that is POSIX system programming, using exciting and not very convincing shark-themed metaphors!

Using rufus-scheduler with Passenger or Unicorn requires a bit more knowledge and tuning, gently provided by a bit of googling and reading

I don't know what this Ruby thing is, where are my Rails? I'll drive you right to the tracks.

And, of course, I'm grateful to be reunited with my long lost friend, false.

but PR is not updated somewhy

It's an interesting position and had me rethinking things a bit, but the way I look at it, the actions themselves are negative; it's their boundary conditions which are different. Take for instance embark/disembark. In pseudo-mathematical terms, I would tend to think they increment or decrement one's embarkedness, with an upper boundary of 1 (aboard), and a lower boundary of 0 (ashore). The non-existence of values >1 (super-aboard) or <0 (anti-aboard) shouldn't affect the relative polarity of the actions themselves. I think. Looking through the rest of the list, there's a variety of different boundary conditions. Prove/disprove would range from 1 to -1 (1=proven, 0=asserted but untested, -1=proven false), entangle/disentangle seems to range from 0 to infinity (because you can always be a little more entangled, can't you?), and please/displease is perhaps wholly unbounded (if we imagine that humanity has an infinite capacity for both suffering and joy).

I recently chatted with a data science leader who described their company reaching this state. They couldn’t show any business impact from the past two years of their product releases, so the finance team identified a surefire way for R&D to make a business impact: laying off much of the R&D team.

2) You will rarely get greenfield projects

when I finish reading an article, I'm excited to go to Tinderbox and play with what I've just learned. And that is just rare. Normally that sort of work is is tedium and it doesn't feel that way.

You don’t need to build your brand. You are allowed to have fun on the internet.

In my experience, client software usually honors Reply-To, but I have only been exposed to a small fraction of the wide array of client software that people might use.

Rick was a very talented developer. Rick could solve complex business logic problems and create sophisticated architectures to support his lofty designs. Rick could not solve the problem of how to work effectively on a team.

I dove into the source code. Rick was right: no-one could possibly understand what Rick had created. Except for Rick. It was a reflection of the workings of his own mind. Some of it was very clever, a lot of it was copy-pasta, it was all very idiosyncratic, and it was not at all documented.

Backpacker Jobs Backpackers and vagabonds do work that I’ll call “alternative” travel jobs. The type of work that may not require a computer or a college degree, but has a more hands-on approach. Think musicians, artists, or manual labor. Pay could be under the table. Examples: Street vendor, musician, farm work, etc.

If you lose the desire to be silly, the power to laugh, and the ability to poke fun at yourself, you will lose the power to think. All work and no play makes Jack a dull boy for one reason: It kills off his imagination.

You just blew my mind Marc-André. My code just went from idiotic to idiomatic.

To take advantage of this wonderful opportunity, put #! /bin/sh at the left margin of the first line of your shell scripts.

We all like games with a wide variety of resources, but most board games or card games never seem to have all that many.  So let's do a comparison with popular games out there.

These outcomes are often achieved in the face of non-trivial levels of defection, free-riding, and other bad-actor patterns.

Vehicular traffic comprises millions of objects, each weighing up to several tons, moving at high speeds in close proximity. Yet traffic is able to flow reasonably safely thanks to a relatively small set of rules, starting with a consensus about which side of the road to drive on.2.

Joseph Lister

Destroys the concurrency of your ruby code using Redis.

“I only dowhat is easy. I only write when I immediately know how to do it. If Ifalter for a moment, I put the matter aside and do something else.”(Luhmann et al., 1987, 154f.)[4]

“I only dowhat is easy. I only write when I immediately know how to do it. If Ifalter for a moment, I put the matter aside and do something else.”(Luhmann et al., 1987, 154f.)[4]

net send * MESSAGE

eierlegende Wollmilchsau

In other words, someone has to clean up after "irresponsible" parents that leave their children un-wait'ed, and that's PID 1's job.

created an ancillary tool, named Yardstick, which verifies YARD (or RDoc) documentation coverage.

Building machines and running downscaled tests is my late-night hobby.

When I find a "get X free" button on a website that then asks for my email address, I like to search for the email of the company behind the website (sometimes it's on the legal page, or the privacy policy page) and I submit their email. I also make sure to check the "sign me up for the newsletter" box, to make sure the spammers get at least one of their messages. I don't really know why I do this, it seemed funny a few months ago when I started and now I do it out of habit. I now keep a list of emails from these spam sites, and subscribe them all to the various newsletters I find if I have 5 minutes.

Assuming a conservative annual growth of digital content creation of 1%, using (3), we estimate that it will take around ∼3150 years to produce the first cumulative 1 kg of digital information mass on the planet and it will take ∼8800 years to convert half of the planet’s mass into digital information mass

The perpetrator in question was completing an internship and committed code into the Windows 3.1 code base that was a little prank for the test team: Under a very specific error condition, it changed the index finger pointer to a middle finger.

All I know for sure is that they are trapped in the box.My solution for them: This isn’t working. Free yourself. Get out of this box. Putit away for another day and start a new box.

Type 2 fun is miserable while it’s happening, but fun in retrospect. It usually begins with the best intentions, and then things get carried away. Riding your bicycle across the country. Doing an ultramarathon. Working out till you puke, and, usually, ice and alpine climbing. Also surely familiar to mothers, at least during childbirth and the dreaded teenage years. I remember that very trip to Alaska, just a week before learning about the Fun Scale, when Scott and I climbed Mt. Huntington. Huntington might be the most beautiful mountain in the Alaska Range, but the final thousand feet was horrifying—steep sugar snow that collapsed beneath our feet as we battled upward, unable to down-climb, and unable to find protection or anchors. On the summit, with the immaculate expanse of the range unfolding in every direction, Scott turned to me and said, in complete seriousness, “I want my mom so bad right now.” By the time we reached Talkeetna his tune changed: “Ya know, that wasn’t so bad. What should we try next year?”

magic

fun, fairness, and challenge

Hardcorps

VEXILLOLOGY: The scientific study of the history, symbolism, and usage of flags or, by extension, any interest in flags in general. — Whitney Smith, Flags Through the Ages and Across the World, 1975

Esrever

It is a perfectly cromulent word. :)

Nice use of “embiggen” as a class name!

It is a descendent of the "Object Mother" pattern for creating objects for testing, and is related to the concept of an "object exemplar" or stereotype.

That's it! Just replace let! with let_it_be. That's equal to the before_all approach but requires less refactoring.

Another interpretation of the “Small Web” concept is that it refers to the use of alternative protocols to the dominant HTTP(S), lightweight ones like the older Gopher and newer Gemini. For example, the blog post Introduction to Gemini describes these collectively as “the Small Internet”.

As you read along, you’ll begin to notice UI designer is just another word for a foodie—we love to name our UI elements after food.

Cape's Escape Game 2.5th Room

µ

#blogumentation

die-hard fans of filter-branch may be interested in filter-lamely (a.k.a. filter-branch-ish), a reimplementation of filter-branch based on filter-repo which is more performant

gilding

GRADE K-4 GRADE 5-6 GRADE 7-8 GRADE 9-12 SPANISH TECH TEACHER Teacher Sign Up Sign In Teacher Sign Up Sign In GRADE K-4 GRADE 5-6 GRADE 7-8 GRADE 9-12 SPANISH TECH TEACHER TT GRADE K-4 GRADE 5-6 GRADE 7-8 GRADE 9-12 SPANISH TECH TEACHER Teacher sign up Sign In Why did ancient Greeks and Romans eat lying down? (Thinkstock) Why did ancient Greeks and Romans eat lying down? By: Ask Smithsonian, Smithsonianmag.com November 25, 2015 Published: November 25, 2015 Lexile: 1230L var addthis_config = { services_exclude: 'print,printfriendly', data_ga_property: 'UA-6457029-1', data_track_clickback: true }; var addthis_share = { url_transforms : { shorten: { twitter: 'bitly' } }, shorteners : { bitly : {} }, templates : { twitter : '{' + '{title}' + '}: {' + '{url}' + '} via @TweenTribune' } }; 530L 780L 1040L 1230L Assign to Google Classroom You asked us, "Why did ancient Greeks and Romans eat lying down?"   Reclining and dining in ancient Greece started at least as early as the 7th century BCE and was later picked up by the Romans.   To eat lying down, while others served you, was a sign of power and luxury enjoyed by the elite. People further down the social ladder copied the laid-back dining style, if they could afford to.   I mean, who wouldn't want to stretch out while chowing down, but not everyone was so lucky in ancient Greece. You see, women didn't generally get invited to banquets except for rare occasions like wedding feasts and even then they had to sit upright.   It was only in ancient Rome that customs changed, allowing upper-class women to lounge alongside men, and while it sounds sweet, all that lying down and eating can't have been good for the heartburn. Source URL: https://www.tweentribune.com/article/teen/why-did-ancient-greeks-and-romans-eat-lying-down/ Filed Under:   Video Culture Odd news Smithsonian Assigned 49 times CRITICAL THINKING QUESTION Why did people “further down the social ladder” copy people above them? Write your answers in the comments section below Please log in to post a comment COMMENTS (15) arellanoj-rob 11/30/2015 - 09:46 a.m. I think that people "further down the social ladder" copied people above them because they thought it'd earn them some sort of respect. It probably gave them sense of power back then. julianc-bag 11/30/2015 - 07:32 p.m. I don;t like eating at the dinner table I prefer the living room. ShawnaWeiser-Ste 12/02/2015 - 03:56 p.m. This seems quite unnecessary and dangerous. Its very common for people lying down to choke while they are eating, I mean come on. Good thing the women and the poor were not allowed to engage in such activities; they probably lived much longer than the rich men. laurenc-bag 12/03/2015 - 09:18 p.m. People "further down the social ladder" copied people above them, possibly to make themselves look a little wealthier than they were. It was a sign of luxury and was only enjoyed by the elite, so they wanted to experience that as well. laurenc-bag 12/03/2015 - 09:21 p.m. And, most likely, my weirdest custom at home is listening to music while watching a video on my phone while FaceTiming my friends, if that even counts as a strange custom... But, I also pray before I eat every meal with my family, which might seem strange to some people. laurenc-bag 12/03/2015 - 09:30 p.m. (It didn't allow me to take the test for some reason...) carsonb-2-bar 12/03/2015 - 10:28 p.m. In the early 7th century reclining and dining in Greece started and later on picked up by the Romans. According to the article it was a sign of power, especially when others served you. People in lower social classes copied it. The lower class people probably copied the upper-class people to be cool. Maybe it made them feel powerful. I thought the article was interesting. I never knew why many pictures back in the 7th century show people eating while lying down. I guess you could say they were the first couch potatoes! bellae1-lin 12/04/2015 - 02:57 p.m. People "further down the social ladder" copied people above them because they wanted to feel luxurious and wealthy. They would want to feel this way because they may not be treated like luxury, and they wanted to see with the eyes of a wealthy being. briannec-ste 12/07/2015 - 05:09 p.m. I personally don't like to eat laying down because I feel like I am being choked. I don't understand how laying down and being fed was a sign of wealth. The laying down not at all but the getting fed I understand. gisellem-pay 12/08/2015 - 11:11 a.m. I think that this concept is similar to our current society. Many people find or develop a custom, in which will catch on to others just to prove their power or how modern they believe they are. This also reminds me of China and foot binding. This tradition was passed down for women as a beauty concept. Page 1 of 2 Next » Take the Quiz Leave a comment ADVERTISEMENT TOPICS Animals Video Education Art Entertainment Culture Food & Health Inspiration National news Odd news Science Technology World news ADVERTISEMENT LEXILE LEVELS 500L-590L 600L-690L 700L-790L 800L-890L 900L-990L 1000L-1090L 1100L-1190L 1200L-1290L 1300L-1600L ADVERTISEMENT Take the Quiz Leave a comment ABOUT US FAQs Terms of Use Privacy Statement LOGIN Sign In Teacher Sign Up Can't Login GET IN TOUCH Contact Us Facebook Twitter Pinterest RSS The Smithsonian Institution is a trust instrumentality of the United States established by an act of Congress in 1846 "for the increase and diffusion of knowledge" googletag.cmd.push(function() { googletag.display('div-gpt-ad-tt-outofpage'); }); window.webtrendsAsyncInit=function(){ var dcs=new Webtrends.dcs().init({ dcsid:"dcs8v0iiladzpxfcn5y7c8cy2_5j6f", domain:"logs1.smithsonian.museum", timezone:-5, i18n:true, fpcdom:".tweentribune.com", plugins:{ } }).track(); }; (function(){ var s=document.createElement("script"); s.async=true; s.src="https://static.media.tweentribune.com/js/webtrends.min.js"; var s2=document.getElementsByTagName("script")[0]; s2.parentNode.insertBefore(s,s2); }()); <img alt="dcsimg" id="dcsimg" width="1" height="1" src="//logs1.smithsonian.museum/dcs8v0iiladzpxfcn5y7c8cy2_5j6f/njs.gif?dcsuri=/nojavascript&amp;WT.js=No&amp;WT.tv=10.4.23&amp;dcssip=www.tweentribune.com"/>

Wehrlegig Games

teaches players how to play while they play