My 6200 Likes & Retweets

Semantic search improves our agent's accuracy across all frontier models, especially in large codebases where grep alone falls short. Learn more about our results and how we trained an embedding model for retrieving code. https://t.co/qL8np1sf8n

After two years of work, we’ve made an AI Scientist that runs for days and makes genuine discoveries. Working with external collaborators, we report seven externally validated discoveries across multiple fields. It is available right now for anyone to use. 1/5 https://t.co/CpWLzM3ZQI

Today, we’re announcing Kosmos, our newest AI Scientist, available to use now. Users estimate Kosmos does 6 months of work in a single day. One run can read 1,500 papers and write 42,000 lines of code. At least 79% of its findings are reproducible. Kosmos has made 7 discoveries… https://t.co/mV6BfmoZPR

Your agent connects to 50 MCP servers with 1,000 tools. It loads every tool definition into context before reading your request. Then it passes a 10,000-row spreadsheet through the model twice to copy data between tools. You just burned 150,000 tokens on overhead. Code… https://t.co/M84ZPEZYhg

Thank you, Mr. President @POTUS, for this opportunity. It will be an honor to serve my country under your leadership. I am also very grateful to @SecDuffy, who skillfully oversees @NASA alongside his many other responsibilities. The support from the space-loving community has… https://t.co/CUpyTXfB14

Introducing Codemaps in @windsurf! powered by SWE-1.5 and Sonnet 4.5 “Your code is your understanding of the problem you’re exploring. So it’s only when you have your code in your head that you really understand the problem.” — @paulg https://t.co/Tyodea1MrN

Our TPUs are headed to space!  Inspired by our history of moonshots, from quantum computing to autonomous driving, Project Suncatcher is exploring how we could one day build scalable ML compute systems in space, harnessing more of the sun’s power (which emits more power than 100… https://t.co/aQhukBAMDp

Introducing GEN-0, our latest 10B+ foundation model for robots ⏱️ built on Harmonic Reasoning, new architecture that can think & act seamlessly 📈 strong scaling laws: more pretraining & model size = better 🌍 unprecedented corpus of 270,000+ hrs of dexterous data Read more 👇 https://t.co/7VXtFXFaG1

Introducing OlmoEarth 🌍, state-of-the-art AI foundation models paired with ready-to-use open infrastructure to turn Earth data into clear, up-to-date insights within hours—not years. https://t.co/FozH4HIGhp

Continuous Autoregressive Language Models Compresses a fixed number of tokens into a vector using an autoencoder, and directly models prediction of next vector. https://t.co/W5YFo4EINE

ig the goal is to be rich enough that curiosity drives you instead of necessity. https://t.co/YLWPBPPWN1

I'm starting to change my thoughts on where AI is best used during the ideation process. Originally, I thought AI was great at the beginning, helping me discover the universe of ideas on a specific topic. Then, I would choose areas to focus on and expand, also using AI. I now…

a man will get into a competitive field, learn how to build apps, post consistently, iterate over ideas, ship a working product that finally finds pmf and then call himself lucky you’re not lucky you’re just extremely hard working and dedicated https://t.co/kvfrTrcZom

The biggest predictor of coding ability is Language Aptitude. Not Math. A study posted in Nature found that numeracy accounts for just 2% of skill variance. Meanwhile, the neural behaviors associated with language accounted for 70% of skill variance. https://t.co/jlyFUded6e

how to research; properly. most people “google and scroll.” real researchers hunt, map, and build. here’s the generalized approach that actually works; in any field: 1. start with a question, not a topic. “how can we make small drones navigate indoors?” is better than “indoor… https://t.co/Qh8t1zzYAI

X Article: https://t.co/EGm0aVmOGj

@yacineMTB if i were like, a sports star or an artist or something, and just really cared about doing a great job at my thing, and was up at 5 am practicing free throws or whatever, that would seem pretty normal right? the first part of openai was unbelievably fun; we did what i believe is…

For the first time in our existence, we possess the means, technology, and, for the moment, the will to establish a permanent human presence beyond Earth. Starship is designed to make this future a reality → https://t.co/dGAZiB4rr3 https://t.co/WsTg44G3oz

Today we’re launching Perplexity Patents, the world’s first AI patent research agent that makes IP intelligence accessible to everyone. Read more about Perplexity Patents in our latest blog: https://t.co/BUZK1vLr2i https://t.co/fy5KT6cYrW

a message to MKBHD https://t.co/ADodKhgkKB https://t.co/2iNwgGkQKO

Hello Thermo World. https://t.co/yO96FRhoZ6

🚨 NEW LABS EXPERIMENT 🚨 Introducing Pomelli, an experimental AI marketing tool designed to help you easily generate scalable, on-brand content to connect with your audience, faster. Just enter your website, and Pomelli will understand your unique business identity to build… https://t.co/rpmk0ud0x2

Substrate is building a next-generation foundry to return America to dominance in semiconductor production. To achieve this, we will use our technology—a new form of advanced X-ray lithography—to power them. America invented semiconductors. We will lead again. https://t.co/1rgnWSYYZG

BREAKING: Chinese influencers now required to hold university degree to discuss serious topics — or face a ¥100,000 fine.

Introducing Pinch to Build A completely new interface for vibe coding professional apps One that makes building mobile apps feel like using CapCut 7 HUGE updates 🧵⬇️ (Reply to this post, & we'll DM you free access right now) https://t.co/r9WV2bPxsV

LLMs are injective and invertible. In our new paper, we show that different prompts always map to different embeddings, and this property can be used to recover input tokens from individual embeddings in latent space. (1/6) https://t.co/HF5Y1c5iWS

Math is overpowered when combined with other expertise. But it's also underpowered alone. https://t.co/Pm9fEClICl

non-technical? no. i am post-technical.

Good reminder on the state of AI https://t.co/xwgdTiAc0D

we are moving from a knowledge economy to an allocation economy https://t.co/UEpES3abqh