On September 22, 2026, Anthropic shipped Claude Opus 5.5 — its first new model since CEO Dario Amodei published an essay ten days earlier arguing the whole industry should deliberately slow down. It's 40 percent cheaper than Opus 5, runs about 30 percent faster, performs close to the pricier Fable 5
Naive AI, founded in Beijing in February 2026 by Tsinghua professor Dai Jifeng, has raised about $400 million across three rounds at a $1.42 billion valuation, backed by Tencent, IDG Capital, MPCi and HSG. It has fewer than a hundred employees and has not released a model. Every one of those investo
Anthropic's own newsletter told me to run a new command, /skill-doctor, to see what my accumulated Claude Code skills were costing me. It failed three times, and once, when it printed the true reason why, the assistant's own reply talked past it. The fourth try stopped refusing the connection and ga
In one week in September 2026: OpenAI shipped a model its president called the start of the AGI era, then asked Congress for mandatory safety rules it had spent two years opposing; it put its most prominent internal critic on the board, and he said the industry was not on track; Anthropic published
For three months I have worked every day with an AI coding assistant that wakes up each session knowing nothing about the last one. What it knows, it reads from a file. That file grew to 685 kilobytes, went stale in ways that reached the real-money side of the project, and turned into the best lesso
Some friends and I have been arguing about whether it will ever seem normal to spend six million tokens in eleven minutes to make a blog post — and whether a world where the well-off can run AI flat out leaves everyone else with the crumbs. So I priced the tokens, traced where the price is going, an