Kimi K2.5 1T runs on 2 M3 Ultras with mlx-lm in it's native precision. It's actually quite usable. Here it's making a space invaders game. Generated 3856 tokens at 21.9 tok/sec using 350GB per machine. Thanks to @kernelpool for the port.

Boom Supersonic CEO @bscholl explains why we got Star Trek's iPads but not its warp drives: "Star Trek Next Generation, touchscreens, everyone’s walking around with an iPad. We got that, but we didn’t get the photon torpedoes, we didn’t get the warp drive. We got all the low energy sci-fi, but we didn’t get the high energy sci-fi." "Thomas Edison flipped the switch on the very first electric grid in 1882. From 1882 through to 1973, there was exponential increase in power production per capita, doubling about every 11 years. And then in 1973, something went wrong, and that trend stopped." "Since the year 2000, our population has increased faster than our energy. The amount of energy available to the average American is actually less now than it was 26 years ago." "We’ve shipped all the manufacturing to Asia. I think we all know now that was a terrible idea. China’s got three times the energy production of the US." "The way to bring it back is not bring it back, but invent the next generation of manufacturing technology here. We’re gonna need a lot of power for that. And of course, AI." @boomsupersonic

There is an (unpatchable?) supply-chain zero-day in PyTorch autograd that allows a regularization term to use the model's weights as RAM and SGD as its CPU clock to run any computation during training.

Pace of Linux kernel development is a clear indication of the acceleration in the amount of software being written! 7.2 is the largest, following 7.1. lwn.net/Articles/1088776/

Codex is one of the most important untold stories in the history of technology. A year ago, it was a scrappy effort in a quiet corner of @OpenAI. Its cloud product hadn’t found PMF, and the team was starting over from scratch. Less than 12 months later: - Codex swallowed the flagship ChatGPT desktop app from the inside (it literally runs as com.openai.codex on your Mac) - Codex and ChatGPT for Work hit 40M+ weekly users across engineering and knowledge work - Its leader @thsottiaux was elevated to run all of OpenAI’s core products, including ChatGPT, Codex, and the API How did they pull it off? They trained their own dedicated model line, separate from ChatGPT, with researchers and product builders sitting on the same team They built four different products in a single year (cloud → terminal → IDE sidebar → a desktop app with no code editor at all) as every model leap made a new form factor possible They used Codex to build Codex—so a handful of people could prove a new product in weeks instead of arguing for it in meetings I had a front-row seat to the whole arc—from their failed Cloud launch, to OpenAI’s offices the night before GPT-5 dropped, to the day Codex took over ChatGPT. And I spent months interviewing @gdb, @thsottiaux, @embirico, @mia_glaese, @aambrosino, and @edbayes to get inside how it all went down. Here is the full, behind-closed-doors story of the team that remade OpenAI: every.to/p/how-codex-happene… We’re pushing the boundaries of how stories like this are told. This piece is completely agent-native—designed to be read alongside your agent inside ChatGPT’s in-app browser: - Every interview transcript is included in full: Click any quote and you (or your agent) can see the exact conversation it came from. - Built for your agent to explore: Ask your agent to pull out product lessons for your team, cross-examine the transcripts, or find insights I missed when writing the piece. It’s a story about the future of software—built to be read with an agent: every.to/p/how-codex-happene…

Also, why was the bug encased in white toilet paper to conceal it on a black TV? What’s the thinking here?

Assorted things I've been mulling over regarding agents in the economy: • Companies are in some sense the original superintelligence. It's often the case that they can use their amortized cognitive surplus to advance their interests against those of certain consumers: relative to the optimization capacity of companies, consumers can be inattentive, underinformed, irrational, etc. (Think, say, of some of the classic insurance upsells at the end of a booking process, which is typically priced substantially above what a regular policy would cost.) Herbert Simon introduced the idea of bounded rationality, and firms often end up benefiting from it. Thanks to personal agents, we will probably get somewhat more rational in how we engage with companies. • How exactly will outcomes change? Well, certain kinds of price discrimination will work less well (how do coupons work if every agent hunts fastidiously for them?), and certain kinds of customer cross-subsidy will change (today, people who forget to cancel subscriptions subsidize those who don't). Taken across the economy as a whole, I'm not sure what the incidence of all of this will be. • In corporate strategy, people have long debated the relative importance of distribution and product quality. Perhaps you've made something better, but does anyone know about it? Won't people just continue to buy the ACME Corp Mousetraps? Personal agents will, I think, act as a kind of structural subsidy for product quality. How would markets look different if individuals always spent at least 10 hours researching their purchases? On the margin, focusing on making a product better is, I think, going to become a more effective strategy for companies. This is good! • It true, a corollary is that information about product quality will become more important and impactful. Today, most people have very limited ability to aggregate and integrate such data, but agents will have insatiable thirst for such signals. (It's often said that data is becoming more valuable because it helps in the training of models. That's obviously true, but I think there's a secondary way in which it's more valuable simply as a complement to cheaper intelligence.) • A big question is whether agents will increase superstar effects (rational agents all settle on the same product) or increase dispersion (because everyone has slightly different tastes and preferences, which agents are good at eliciting, finally overcoming the artificial clustering we've been subjected to). • Many internet companies have monetized demand routing, which is in turn often predicated on short attention spans. "I'll click on the first thing I see." This has led to some amount of inefficiency, as companies try to propitiate capricious ranking algorithms and incur various placement fees. It's not all deadweight loss (willingness to pay is itself a kind of signal, which is valuable to buyers), but it's not great. How will the nature and value of this routing change in a world of personal agents? I don't know the answer, but I think that the effects might be pretty big. • Much of what I'm writing hinges on the assumption that personal agents will be on the consumer's side. I suspect that that will be the winning strategy, but whether it's true, and what it means to be true, involves a lot of other questions. Taken as a whole, the market is a massively parallelized, decentralized, and highly imperfect value attribution machine, and, taking stock of all of the effects, it seems to me that the quality point may well be the biggest: that agents will change the nature of the backpropagation such that the rewards for better products increase. Maybe this is too optimistic, but I suspect (and hope) that companies that make things that they're proud of will feel that the universe is a little more partisan in their favor.

Hearing so many of these stories.

the choice of whether to resist the future or rapidly adapt. some strategic wisdom from our partner @mvernal

We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks: Writing. Something I've had success with: Ask your LLM to explain something in ASD-STE100, it's a controlled language specification originally developed for aerospace maintenance documentation. LLMs well-versed in this language and it comes with heavy constraints on clean writing style that I often find a lot more readable. Sometimes I've tried to soften it a bit e.g. ask for "80% of the way to ASD-STE100" because the spec is quite stringent. But even better: Diagrams / images. Instead of writing, ask your LLM to create a diagram. These can be a lot easier to process, parse, and understand. But even better: Web pages. Ask for output "in HTML" to get a beautiful, interactive webpage. LLMs are getting really good at frontend and can create beautiful experiences, animations, etc. But even better: Explainer videos. The output format I am most bullish on is fully custom / bespoke explainer videos generated on any arbitrary topic. Experiment with things like "Create a 3b1b style video explainer on X. Use my ElevenLabs API key for audio narration". (you'd need an API key for the latter or you can ask your LLM to find you decent free alternatives that use your local compute). This is actually starting to work! In summary: - As LLMs get better, they will do more and more of the legwork autonomously, and a lot more of our work will rise up the abstractions into oversight and understanding. - Luckily, LLMs can help here too because as intelligence and code are increasingly abundant, you can ask for large, custom, discardable software artifacts (e.g. web apps, video explainers) that would have never made sense to create before. Push the boundaries here and you'll be surprised.

Took a minute to write a few words about security & safety as someone who lived through it all at OpenAI. I hope my thoughts help someone out there.