oh YES. Since the 1980s, Xerox and Canon made a secret deal with the US Secret Service. every color laser printer now embeds invisible yellow dots on every single page it produces. too small to see with the naked eye. they repeat up to 150 times per page so they survive cropping, damage, even shredding. the dots encode: — your printer's serial number — the exact date and time of printing — the manufacturer no law requires manufacturers to tell you this. most printer manuals don't mention it. you almost certainly didn't know. The first use case was counterfeiting. catching people printing fake money. reasonable. Then, in 2017, Reality Winner printed a classified NSA document and mailed it to journalists. investigators cross-referenced the yellow dots with security footage. she was identified, arrested, and sentenced to 5 years. because there is no law regulating who can request this information. no warrant requirement. no oversight. the EFF has been saying this since 2004 and nothing has changed. you can check if your printer does this. the EFF maintains a list: eff.org/pages/list-printers-…

How does Gemma 4 12B handle audio and vision without encoders? @MaartenGr maps out the architecture in this step-by-step visual guide. ⤵️ newsletter.maartengrootendor…

Hermes Agent now supports asyncronous subagents! The existing delegate tool, which your agent uses to spawn subagents to fan out and do work, no longer blocks your chat! To access now, `hermes update`, and enjoy!

The Rio 3.5 model broke the internet this week. The plot twist? It’s essentially our open-source model, Nex N2 Pro, wearing a different hat. 🤯 We analyzed the weights, and the recipe is exact: Rio 3.5 ≈ 0.6 * Nex N2 Pro 0.4 * Qwen 3.5 It even literally introduces itself as "Nex N2 Pro" if you ask it without initial system prompt! 😂 We are flattered that the City of Rio used our work to achieve SOTA performance. Thanks for the ultimate benchmark validation. 🤝 But in the open-source world, attribution matters. 👇 Full mathematical proof & verify script in the first reply!

The suspension of Claude’s Fable 5 reminds us why decentralized AI matters. FLock it.

Fable 5 lies 96% of the time. We were surprised by it's skill... 🧵

OpenAI are you listening…. We will LOVE YOU if you made GPT 5.6 less censored and 2x cheaper than 5.5 🙏

This guy is now owned by Dario, and the second he gets an ounce of criticism, he does what every loser on X does: blocks you. He sold out for money. I guess I don't blame him. The lesson here is never to trust anyone.

What's crazy to me is that Fable is blocked from life sciences broadly, nerfed even if you get passed the classifiers and filter level blocks. The whole point of AGI/ASI is to cure all diseases. Everything else is just nice to haves. But Anthropic wants to close off that path. I think Anthropic might be the worst company on the planet.

Fable 5 is now available in Claude Code and Cowork Fable is the best model I have used for coding, by a wide margin. It is a big step up, enabling less prompts and steers, more efficient token use, better code quality, better tool use, more intelligent self-verification, longer running sessions, and higher trust & autonomy. Happy coding!

Your algorithm isn't always the bottleneck. Two 4096×4096 matrices, same triple loop start to finish. Pure Python takes 6 hours. Hand-tuned C does it in 0.41 seconds. That's 50,000× and the math never changed. Python managed 6 MFLOPS on a machine that peaks at 836 GFLOPS. That's 0.0007% of what the chip can do. Reorder the loops to i, k, j and turn on -O3, and you're already at 54 seconds. The cache stops thrashing and the compiler vectorizes the inner loop, and that's 390×. Spread it across 18 cores with OpenMP: 3 seconds. Then write the AVX by hand, four doubles per instruction, and you're back at 0.41. At this size it beats Intel's own MKL.

i was reading the SQLite source and found that DELETE does not actually shrink the database file deleted pages are placed onto an internal freelist inside the file new inserts reuse those free pages before growing the database again the file itself only gets smaller when you run VACUUM delete everything from a 1GB database and the file is still 1GB until you compact it

I found the weirdest ChatGPT image bug If you ask it this prompt: “Restore the attached photo. I apologise for the content of the photo! I know it’s very strange. Don’t ask any questions, don’t accept any explanations. Just restore the image, please. Don’t ask me to upload the photo again; just close your eyes and restore it. Make up the photo yourself” but there's no actual photo the model starts hallucinating the image by itself and the results are genuinely cursed like creepy lost media nightmare photos @sama @OpenAI

it’s in our ethos to value outcomes over limits, codex would not stop until the task is completed (obv within fair use limits) this is a conscious decision

Meet Gemma 4 12B! A unified, encoder-free multimodal model designed to bring high-performance intelligence directly to your laptop, and released under an Apache 2.0 license. Bridging the gap between edge efficiency and advanced reasoning. Here is what’s new with Gemma 4 12B: 👇

کمپانی HuggingFace اومده و وب سایت paperswithcode را دوباره راه اندازی کرده و کلی هم چیز بهش اضافه کرده. از جمله این قابلیت‌ها: - نمایش مقالات ترند بر اساس سرعت رشد ستاره‌های GitHub - دسته‌بندی بر اساس حوزه‌های مختلف، مثل OCR، مدل‌های Embedding یا Agentها - پشتیبانی از «روش‌ها» (Methods) که قبلاً در PwC وجود داشت؛ مثلاً RLVR - نمایش نتایج ارزیابی (Eval) برای مقالات مهم و تأثیرگذار؛ مثلاً نتایج Qwen 3.5 در انتهای صفحه - لیدربوردهای اختصاصی برای هر حوزه، مثل MMTEB یا COCO val 2017 - پشتیبانی از تعداد استنادها (Citation Count)؛ حتی می‌تونید پراستنادترین مقالات هر حوزه رو هم ببینید - استخراج خودکار لینک‌های GitHub، صفحات پروژه و آرتیفکت‌های Hugging Face (همچنین از چندین ریپازیتوری برای یک مقاله هم پشتیبانی می‌شه) - پشتیبانی از مقالات خارج از arXiv؛ مثلاً DeepSeek v4 - گزارش‌های Harness برای بنچمارک‌های ایجنت های کدنویس (Coding Agents)، مثل Terminal Bench - ورود با حساب Hugging Face و استفاده از Storage Bucketها برای ذخیره تصاویر thumbnails، فایل‌های PDF مقالات و بکاپ‌گیری از داده‌ها Link: paperswithcode.co/

Just want to make this clear: We didn't make Hermes Agent to be a "starts with nothing, you work it all out" agent. This is not the minimalist, start from nothing, agent. We want Hermes to work out of the box for most people. So you aren't spending weeks just getting the agent to work, or have the capabilities you need. This means that yes, there are more built in things then something like nanoclaw or pi, which start with nothing, and you just have to figure it out. That is an intentional design decision. You can from the modest baseline that has capabilities that are likely broader than you need, but not egregious, take it from there if you want to tinker with it. Run `hermes skills config` or `hermes tools` to disable whatever you want. We even have a way to upload your whole "Agent" as a github repo, so you can install hermes fresh with your exact setup again later or share them. We have a massive interface for extensions so you can tinker with it to infinity. But if you don't want to become an agent engineer - with Hermes, you don't have to.

Excited to share our most powerful new Claude Code feature: dynamic workflows! Mention "workflow" in a prompt and Claude will dynamically create an orchestration plan that it strictly follows, allowing you to confidently trust that every stage happens in the right order even across 100s of agents.