Here’s something unexpected: Google’s latest Gemini 3 model can draw a pelican riding a bicycle—and it’s actually good. This seemingly trivial benchmark reveals something profound about where AI image generation has arrived. The Core Insight...
A 9,649-experiment study challenges conventional wisdom about how to feed AI agents structured data How should you structure the context that AI agents consume? The answer might surprise you: it depends on your model, and...
China’s largest open-weight AI model raises the stakes — and coins a new term for AI-era software development If you thought the AI model size wars were cooling down, think again. Z.ai just released GLM-5,...
How tree-sitter indexing and the Recursive Language Model pattern could transform how AI understands code What if AI agents could navigate your codebase the way an experienced developer does — not by reading everything, but...
OpenAI’s new ultra-fast coding model hits 1,000 tokens/second — and changes how we think about AI-assisted development What if the biggest breakthrough in AI coding isn’t about quality — it’s about speed? That’s the provocative...
How a nonprofit dedicated to “benefiting humanity as a whole” quietly dropped safety language and financial constraints In 2016, OpenAI debuted with a bold, almost idealistic mission: advance digital intelligence in a way that benefits...
While OpenAI was quietly editing its mission statement, Anthropic took a different approach. Let’s look at what we can learn from both. The Core Insight Following Simon Willison’s analysis of OpenAI’s mission evolution through IRS...
Simon Willison just released two new tools that tackle one of the most pressing problems in AI-assisted coding: how do you know what your agent actually built? The Core Insight When you’re working with coding...
What can you learn about an AI company from its IRS filings? More than you’d think. The Core Insight Simon Willison did something clever: he dug through OpenAI’s nonprofit tax filings from 2016 to 2024,...
What if the biggest breakthrough in AI coding isn’t quality—it’s latency? The Core Insight OpenAI’s new GPT-5.3-Codex-Spark isn’t the best coding model they’ve ever built. It’s not even close—the pelican on a bicycle benchmark shows...