A beautiful book by Google DeepMind, explaining everything you need to know to build your own LLM.

Waited almost three weeks to start scaling LLMs with this one

vik For well executed reasoning RL I would say: arxiv.org/abs/2505.22312 arxiv.org/abs/2506.13284 arxiv.org/abs/2508.06471 arxiv.org/abs/2504.13914 arxiv.org/abs/2508.08221 arxiv.org/abs/2505.08311 arxiv.org/abs/2506.13585 github.com/Tencent-Hunyua… honorable-payment-890.notion.site/POLARIS-A-POst…

We spent the last year evaluating agents for HAL. My biggest learning: We live in the Windows 95 era of agent evaluation.

The 3rd edition of my book Deep Learning with Python is being printed right now, and will be in bookstores within 2 weeks. You can order it now from Amazon or from Manning. This time, we're also releasing the whole thing as a 100% free website. I don't care if it reduces book

Announcing our public preview of Chrome DevTools MCP! Experience the full power of DevTools in your AI coding agent → goo.gle/4pDE6Tk With Chrome DevTools MCP, your AI agent can run performance traces, inspect the DOM, & perform real-time debugging of your web pages.

For all London based founders, partnering with Google and hosting a mixer to celebrate our launch. Oct 15th 5:30PM at the London AI Hub. Luma invite below.

You can now ship ChatGPT apps on Vercel. vercel.com/changelog/chat…

Most "agents" today are shallow while loops around an LLM and tools. Works for simple tasks, but fails when a task requires 50+ steps over several days. To solve complex problems, we need an architectural shift towards Deep Agents (Agents 2.0). We have to decouple planning

Excited to release new repo: nanochat! (it's among the most unhinged I've written). Unlike my earlier similar repo nanoGPT which only covered pretraining, nanochat is a minimal, from scratch, full-stack training/inference pipeline of a simple ChatGPT clone in a single,

What if scaling the context windows of frontier LLMs is much easier than it sounds? We’re excited to share our work on Recursive Language Models (RLMs). A new inference strategy where LLMs can decompose and recursively interact with input prompts of seemingly unbounded length,

What is the origin of this meme 🤣

xAI Hackathon – the ultimate arena for the most hardcore product builders. 24 hours & exclusive access to upcoming Grok models & X APIs Apply by 11/22 🔗 x.ai/hackathon

In this new essay, I lay out the framework for what truly spatially intelligent world models must achieve—three essential capabilities for building AI that creates with a storyteller's imagination, navigates with a first responder's fluency, and reasons about space with

I have a HUGE product announcement! 🎉 Today, I'm launching Teda.dev, an AI app builder that turns your ideas into working apps in minutes. I’ve spent all my nights and weekends in the past month building this, and seeing it come to life still gives me chills.

We're booking out a cafe on December 10th in London Grab coffee, Cursor credits, co-work, and meet the team Let me know if you'd like to come by

SIMA 2 is our most capable AI agent for virtual 3D worlds. 👾🌐 Powered by Gemini, it goes beyond following basic instructions to think, understand, and take actions in interactive environments – meaning you can talk to it through text, voice, or even images. Here’s how 🧵

finally made it out of BA + in SF until next friday 🌁 hosting a little evening for founders on weds with Build Club and Homebrew Club - dropping the link here. will be: > 5 min lightning talks (open mic) > drinks + snacks > holiday-ish vibes luma.com/bnxkudhu

A “unified inbox” + AI agents is here

If you've ever seen someone tweet some cool shader and thought "I don't really even know what a shader is and at this point I'm too afraid to ask" - I've written something just for you. makingsoftware.com/chapters/shade…