Keshav

This is actually brilliant. Sebastian Raschka just dropped an LLM Architecture Gallery. It is a visual reference collecting architecture figures from The Big LLM Architecture Comparison - covering Llama 3, DeepSeek V3/R1, Gemma 3, Qwen3, Mistral, and dozens more. Each model gets a fact sheet: - Scale (params) - Decoder type (dense vs MoE) - Attention mechanism - Key architectural details Basically a cheat sheet for understanding how modern LLMs actually work under the hood. You can even get it as a physical poster (14570 x 12490 pixels) sebastianraschka.com/llm-architectu…

@elonmusk claude will still be better 🥳

By this weekend, xAI will have three Grok Build models in training simultaneously

Apple just dropped AirPods Max 2. No event. No hype cycle. Just a press release. But if you look past the audio specs, something interesting is happening. These headphones now run Live Translation through Apple Intelligence. Someone speaks to you in Japanese, you hear English. Real time. No phone screen needed. They also shipped Adaptive Audio, where the headphones read your environment and decide how much noise to block. Conversation Awareness detects when you start talking and adjusts automatically. Most companies are selling AI. Apple is hiding it. And I think that's the smarter bet long term.

Dario Amodei just went public with something big. Anthropic has been quietly deploying Claude to the Pentagon and intelligence community for years. Now the DoD is demanding they remove safeguards around two things: 1. Mass domestic surveillance 2. Fully autonomous weapons Amodei's response: "We cannot in good conscience accede to their request." This is wild. The DoD is threatening to designate Anthropic a "supply chain risk" - a label for US adversaries, never applied to an American company. The kicker: these two exceptions aren't even barriers to current operations. They're about what Anthropic WON'T build, not what they won't let the DoD use. Mass surveillance + AI that selects targets without humans? That's not defense. That's something else entirely. ht @anthropic

This is wild. Someone built a working rocket for $96. Not a model rocket. A guided missile that actually recalculates its trajectory mid-flight using a $5 sensor. The MPU6050 (gyro + accelerometer) tells the flight computer where it is, and the rocket adjusts its fins in real-time to stay on course. Total bill of materials: $96. The project is open source on GitHub. ESP32 flight computer, 3D-printed body, folding fins. It actually flies. The barrier to aerospace is collapsing fast. github.com/novatic14/MANP…

An AI facial recognition error ruined this grandmother's life. Angela Lipps spent nearly 6 months in jail. She lost her home. Her car. Her dog. All because AI said she was a bank fraud suspect. She's never been to North Dakota. Never flown on an airplane. Here's what happened: Fargo police were investigating bank fraud. They had surveillance video of a woman using a fake military ID to withdraw thousands. They ran the video through facial recognition. AI matched the woman to Angela Lipps. But it was wrong. The real culprit? Still out there. Lipps was arrested at gunpoint while babysitting her grandkids. Held for 4 months without bail as a fugitive. She only got out because records proved she was in Tennessee the whole time. This is what happens when we let AI make life-or-death decisions without human oversight. No apology. No compensation yet. She's still fighting to get her life back.

Can YOUR machine run AI locally? There is a site for that. CanIRun.ai analyzes your GPU, CPU, and RAM to show you exactly which models will work. From 0.5GB models on a Raspberry Pi to 512GB monsters. canirun.ai

A 60-year-old developer posted on Hacker News: I'm ready to retire. But Claude Code is giving me that same energy and drive I had back in the ASP and VB6 days. It kept me up nights trying to absorb it all. Now I'm chasing the midnight hour again. The thread is filled with devs 40, 50, 60+ saying the same thing: AI coding tools brought back the spark they thought was gone forever. One commenter put it perfectly: I thought those days were gone forever. I was so shocked when I found out that I could experience that feeling again. This is the story nobody's talking about. Not the job displacement fear - the rediscovery. AI didn't just change coding. It brought people back to loving coding.

OpenAI just dropped GPT-5.4. And it's a big one. The new model can now use computers natively - meaning it can actually operate your computer, click around, and handle complex workflows without constant hand-holding. On tasks that would normally require a junior analyst (spreadsheets, presentations, the works), GPT-5.4 matched or exceeded human professionals in 83% of tests. Up from 70.9% with GPT-5.2. It also beats human performance on OSWorld (desktop tasks) at 75% vs 72.4%. And hallucinations? Down 33% compared to GPT-5.2. The model is available in ChatGPT, the API, and Codex. If you build agents or automation tools, this is worth a look.

Donald Knuth just wrote a paper about Claude. For those who don't know: Knuth is the legendary Stanford professor who wrote "The Art of Computer Programming" - considered the Bible of algorithms. He's 87 years old. Now he's analyzing Claude's behavior through the lens of his famous "dancing links" algorithm. The paper is called "Claude's Cycles" and it's blowing up. This is kind of amazing: one of the founding fathers of computer science is actively studying and writing about modern AI.

Lafda Legal and Partners. Now hiring.

Meta AI smart glasses are raising data privacy concerns. With 983 points on Hacker News, people are worried about what data these glasses are collecting and how it is being used.

Google just released WebMCP. This is a new standard for exposing structured tools to AI agents. Agents can now actually take action on websites, not just read pages. Booking, tickets, checkout, all through AI. This is a big deal for agentic AI.

Secretary of War Pete Hegseth just designated Anthropic a supply chain risk. This is unprecedented. The first time an American company has received this labelhistorically for US adversaries only. Anthropics response: no amount of intimidation will change our position. We will challenge in court. This is a big deal.

Dear @ICC, this needs to stop. You're blasting stadium music between EVERY SINGLE BALL in T20 World Cup matches. Not between overs. Between DELIVERIES. Think about what's happening between each ball. >A bowler resetting his plan. >A batter reading the field. >A captain adjusting. That 15-20 second window is where cricket's chess match happens. And you're drowning it in DJ tracks at max volume. Meanwhile at Wimbledon, spectators must maintain complete silence during play. You get EJECTED for making noise during a serve. Phones on silent. No singing. No chanting. Movement restricted between games. Why? Because they understand that an athlete's concentration IS the sport. A death over yorker needs the same mental precision as a match-point serve. Why does tennis protect its players' focus while cricket actively destroys it? Nobody's asking to kill the atmosphere. Cricket fans are the atmosphere. The roar after a six, 80,000 people gasping at a caught-behind, the chants building organically from the stands. THAT is what makes cricket electric. Not a PA system force-feeding songs between every delivery like we're at a wedding DJ night. #T20WorldCup #Cricket #ICC #RespectThePlayers

Dario Amodei just made a bold move. Anthropic is refusing DoD demands to remove safety guardrails for mass surveillance and autonomous weapons. He could have taken the $500M+ in contracts. Instead, he chose principle. This is what AI ethics looks like in practice.

I need to talk about something that genuinely messed with my head today. Boris Cherny built Claude Code. Like, he's the guy. And someone shared his actual daily workflow and I had to read it twice because I thought I was missing something. He runs 10 to 15 Claude sessions at the same time. Every day. Five in his terminal, five to ten on the web, all writing code in parallel while most of us are sitting here waiting for one response to finish. But the thing that really got me was the CLAUDE.md file. Every time Claude screws something up on his team, they don't just fix it and move on. They add a rule to the file. And they tell Claude: "Update your CLAUDE.md so you don't make that mistake again." So Claude is literally writing its own rules. Learning your codebase. Getting sharper every single session. The longer you use it the less you have to explain. That's not a chatbot, that's something that actually compounds. Oh and also he hasn't written SQL in over six months. Claude just pulls the BigQuery data straight from the CLI. He doesn't even think about it anymore. And here's the number that I can't shake: Claude Code now accounts for 4% of all public GitHub commits. Four percent. That's not a trend, that's already happened. I'm not trying to hype anything. I just think a lot of people are still using this thing like a fancy search engine and there's a completely different way to use it sitting right there. The CLAUDE.md template is free. Drop it in your project. Seriously.

He found out about OpenClaw. Let's bomb this stuff 😂

Google Chrome had been crashing on me for 4-5 weeks. I watched every YouTube video, tried every fix. Nothing worked. Last night I installed Claude Code, gave it access to my User folder, and typed one prompt: "Google Chrome keeps crashing and I can't figure out why. Figure it out and fix it." It diagnosed the problem in under a minute: 1.5 GB of bloated cache plus GPU memory pressure was killing it. Then without me lifting a finger, it: >Closed Chrome >Cleared 1.5 GB of cached data >Wiped the GPU, shader, and service worker caches >Disabled hardware acceleration >Enabled Memory Saver mode to stop inactive tabs from eating RAM >Relaunched Chrome Total time: 90 seconds. Chrome hasn't crashed since.

The creator of OpenClaw just replied to one of my tweets!! I had called out 800+ malicious skills on Clawhub where the website labels something as safe while the CLI flags it as suspicious. Pretty bad look for a platform where developers download and execute code. Most founders would go silent or get defensive. Peter Steinberger just showed up publicly acknowledged the issue and said he'd fix it. No spin or corporate nonsense. Just honesty. Building open source developer tools is brutal. You ship something useful and the thank you is strangers tearing it apart. Most people would quit. Peter keeps building. Respect to founders who take public heat and channel it into making the product better instead of turning it into a PR crisis. The foundation is clearly solid. The intent is good. That already puts OpenClaw ahead of most platforms out there. Looking forward to seeing the security layer get tighter. @steipete keep shipping