2026-04-16
Pi extension for async subagent delegation with truncation, artifacts, and session sharing
Control your Mac with detailed mouse, keyboard, screen, and window management capabilities.
TPU inference for vLLM, with unified JAX and PyTorch support.
llm201n: neural networks zero to super hero. the bridge from mirograd to tinygrad!
Rendered markdown + LaTeX preview for pi
A claude.ai feature I really like is you can tell it to "clone x/y from GitHub" and it can then answer questions about a repo, or use snippets of code from that repo to help build new artifacts - used that just now to solve a minor friction simonwillison.net/2026/Apr/16/...
Last summer, I posted a thread and screenshots of ny replay of Castles of Dr. Creep on a #c64 emulator. This year I am replaying the sequel: Dungeons of Dr. Creep on my #commodore64 Ultimate. Tonight I finished a really hard level called Dirty Tricks! 🙌😁 Here is a thread and a few shorts 👇👇👇
Shocking result on my pelican benchmark this morning, I got a better pelican from a 21GB local Qwen3.6-35B-A3B running on my laptop than I did from the new Opus 4.7! simonwillison.net/2026/Apr/16/... Qwen on the left, Opus on the right
Apropos nothing, Sir Ben Kingsley is a living treasure who also happened to sign in to some epically bad movies like Species.
The person who attacked Altman’s home referenced AI doomer philosophies as motivation for the attack
I hope @bsky.app gets their rogue “reginos” under control (whatever a “regino” is, I cannot keep up with all these new fancy tech terms)
New video! Talking through my 10+ open model pieces from early 2026 and how they fit together. They're all trying to figure out where open models go next. www.youtube.com/watch?v=hKIX...
Opus 4.7 has a new tokenizer. This means it's also a new base model. Glory days of pretraining still very much going.
The current pace of token-efficient reasoning improvements across minor Claude Opus/GPT model versions is pretty wild. All signs point to this continuing. 4.6 to 4.7 could've been presented as a fairly large model bump in the past with this plot.
I think the adaptive thinking requirement in the new Claude Opus 4.7 is bad in the ways that all AI effort routers are bad, but magnified by the fact that there is no manual override like in ChatGPT. It regularly decides that non-math/code stuff is "low effort" & produces worse results.
I have found that asking for a sestina regularly triggers Opus 4.7's safety guardrails. The forbidden poetic form!
Claude remains irreducibly Claude, across many generations. If you know, you know. (The fact that models have distinct personalities that are consistent across generations is technically interesting, it also makes it easy to use new releases when they come along, because they feel very similar).
Curious to hear from other podcasters -- do you see variation in downloads by day of the week the ep is released? For MAIHT3k, the sweet spot seems to be Tuesday, and I'm not sure why.
When people ask me about "AI" policy, I like to point out that there is policy being made at many different levels and even if our national government is currently a shitshow, local action really matters, like zoning against data centers and "AI" policy in schools.
A "pause AI" letter that I actually like (and signed)! Because it's not about imagined doomsday scenarios, but rather putting the brakes on the rush to impose synthetic text extruding machines in schools in particular: actionnetwork.org/petitions/ca...
<p>For anyone who has been (inadvisably) taking my <a href="https://simonwillison.net/tags/pelican-riding-a-bicycle/">pelican riding a bicycle benchmark</a> seriously as a robust way to test models, here are pelicans from this morning's two big model releases - <a...