Blatant AI slop just won a 25k USD DeepMind Kaggle Grand Prize

Design high-quality benchmarks that go beyond recall to evaluate how frontier models truly reason, act, and judge.

An Engineer's Guide to USB Typе-С (2024)

Solod: Go can be a better C

$100 AI Music Video: Claude Fable 5 vs. GPT-5.6 Sol

We gave Claude Fable 5 and GPT-5.6 Sol the same song, a budget, web search, and local ffmpeg, then let each autonomously direct a music video.

Decoy Font

Decoy Font combines every letter with a decoy, making it harder for AI to read what you type. Type a message to try it out or download the TTF directly.

LM Studio Bionic: the AI agent for open models

The AI agent made for open models, built to get things done.

How Has Roman Concrete Lasted for Millennia? 1,900-Year-Old Latrine Offers Clues

A chemical process called carbonation, which helps seal cracks, could help explain why many ancient Roman structures are still standing today. Researchers hope that the insights will lead to better modern-day building materials

Turn your singing voice into printable notes (in the browser)

Immersive Linear Algebra Book with Interactive Figures (2015)

Pebble Mega Update – July 2026

Pebble Mega Update - July 2026

Camera Chase Vehicle

Starlink from 1984

In good old times, StarLink was a board that turned an IBM PC into a multiuser system.

Microsoft Comic Chat is now open source

Microsoft Comic Chat is now open source. Explore the IRC client that turned conversations into comics and helped introduce Comic Sans to the world.

I Owe My Life to the Commodore 64

If personal computers are for everybody, how come they’re priced for nobody?

Apple targets dozens of OpenAI employees with legal letters

The Little Book of Reinforcement Learning

The Little Book of Reinforcement Learning. Contribute to alxndrTL/little-book-rl development by creating an account on GitHub.

NotebookLM is now Gemini Notebook

NotebookLM is now Gemini Notebook: the same standalone product with deeper Google integration and a secure cloud computer.

Kimi K3: Open Frontier Intelligence

Kimi K3 is the world's first open 3T-class model — frontier performance across coding, knowledge work, and reasoning, with native multimodality and 1M context.

EEG shows brain can simultaneous encode two speech streams

How the brain rapidly switches attention between competing speech streams in complex listening environments is not fully understood. This study shows that neural tracking of a new speaker emerges before disengagement, with transient dual encoding and reduced alpha power supporting flexible auditory attention.