llm-anthropic 0.25
Release: llm-anthropic 0.25 (https://github.com/simonw/llm-anthropic/releases/tag/0.25) • New model: claude-opus-4.7, which supports thinking_effort: xhigh. #66 • New thinking_display and
Simon Willison's Weblog
Qwen3.6-35B-A3B on my laptop drew me a better pelican than Claude Opus 4.7
For anyone who has been taking my pelican riding a bicycle benchmark (https://simonwillison.net/tags/pelican-riding-a-bicycle/) seriously as a robust way to test models, here are pelicans from this
datasette.io news preview
Tool: datasette.io news preview (https://tools.simonwillison.net/datasette-io-preview) The datasette.io (https://datasette.io/) website has a news section built from this news.yaml
datasette-export-database 0.3a1
Release: datasette-export-database 0.3a1 (https://github.com/datasette/datasette-export-database/releases/tag/0.3a1) This plugin was using the ds_csrftoken cookie as part of a custom signed URL,
datasette 1.0a27
Release: datasette 1.0a27 (https://github.com/simonw/datasette/releases/tag/1.0a27) Two major changes in this new Datasette alpha. I covered the first of those in detail yesterday
Gemini 3.1 Flash TTS
Gemini 3.1 Flash TTS (https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-flash-tts/) Google released Gemini 3.1 Flash TTS today, a new text-to-speech model that can
Quoting Kyle Kingsbury
I think we will see some people employed (though perhaps not explicitly) as meat shields: people who are accountable for ML systems under their supervision. The accountability may be purely internal,
Zig 0.16.0 release notes: "Juicy Main"
Zig 0.16.0 release notes: "Juicy Main" (https://ziglang.org/download/0.16.0/release-notes.html#Juicy-Main) Zig has really good release notes - comprehensive, detailed, and with relevant usage
datasette PR #2689: Replace token-based CSRF with Sec-Fetch-Site header protection
datasette PR #2689: Replace token-based CSRF with Sec-Fetch-Site header protection (https://github.com/simonw/datasette/pull/2689) Datasette has long protected against CSRF attacks using CSRF
Trusted access for the next era of cyber defense
Trusted access for the next era of cyber defense (https://openai.com/index/scaling-trusted-access-for-cyber-defense/) OpenAI's answer to Claude Mythos
Cybersecurity Looks Like Proof of Work Now
Cybersecurity Looks Like Proof of Work Now (https://www.dbreunig.com/2026/04/14/cybersecurity-is-proof-of-work-now.html) The UK's AI Safety Institute recently published Our evaluation of Claude
Quoting Steve Yegge
The TL;DR is that Google engineering appears to have the same AI adoption footprint as John Deere, the tractor company. Most of the industry has the same internal adoption curve: 20% agentic power
Exploring the new `servo` crate
Research: Exploring the new `servo` crate (https://github.com/simonw/research/tree/main/servo-crate-exploration#readme) In Servo is now available on crates.io
Quoting Bryan Cantrill
The problem is that LLMs inherently lack the virtue of laziness. Work costs nothing to an LLM. LLMs do not feel a need to optimize for their own (or anyone's) future time, and will happily dump more
Gemma 4 audio with MLX
Thanks to a tip from Rahim Nathwani (https://twitter.com/RahimNathwani/status/2039961945613209852), here's a uv run recipe for transcribing an audio file on macOS using the 10.28 GB Gemma 4 E2B model
SQLite 3.53.0
SQLite 3.53.0 (https://sqlite.org/releaselog/3_53_0.html) SQLite 3.52.0 was withdrawn so this is a pretty big release with a whole lot of accumulated user-facing and internal improvements. Some that
SQLite Query Result Formatter Demo
Tool: SQLite Query Result Formatter Demo (https://tools.simonwillison.net/sqlite-qrf) See my notes on SQLite 3.53.0 (https://simonwillison.net/2026/Apr/11/sqlite/). This playground provides a UI
Kākāpō parrots
Lenny posted (https://twitter.com/lennysan/status/2042615413494939943) another snippet from our 1 hour 40 minute podcast recording (https://simonwillison.net/2026/Apr/2/lennys-podcast/) and it's
ChatGPT voice mode is a weaker model
I think it's non-obvious to many people that the OpenAI voice mode runs on a much older, much weaker model - it feels like the AI that you can talk to should be the smartest AI but it really isn't.
GitHub Repo Size
Tool: GitHub Repo Size (https://tools.simonwillison.net/github-repo-size) GitHub doesn't tell you the repo size in the UI, but it's available in the CORS-friendly API