Some notes on GPT-5.1, which is now available in the OpenAI API The new reasoning options are interesting, but the pelican feels like a bit of a regression from GPT-5 https://simonwillison.net/2025/Nov/13/gpt-51/

Since this question shows up so often that it qualifies as an FAQ, here's my definite answer to "What happens if AI labs train for pelicans riding bicycles?" https://simonwillison.net/2025/Nov/13/training-for-pelicans-riding-bicycles/

I've now reached the "six coding agents in six terminal windows at once" phase of parallel agent delirium https://simonwillison.net/2025/Nov/11/six-coding-agents-at-once/

Sent out my weekly-ish newsletter - I left it a bit too long this time and it ended up with three full articles, two YouTube videos, 5 SVGs of pelicans riding bicycles and 3 POV-Ray renders of pelicans riding bicycles https://simonw.substack.com/p/code-research-projects-with-async

This is really fun - @beetle_b ran a variant of my pelican riding a bicycle SVG test using POV-Ray instead and got terrible ray-traced pelicans out of a bunch of different models: https://blog.nawaz.org/posts/2025/Oct/pelican-on-a-bike-raytracer-edition/ and […]

OpenAI partially released a new model yesterday called GPT-5-Codex-Mini No API access yet, but I did some truly horrible things to their Codex CLI app to get it to spit out this SVG of a pelican riding a bicycle

I have a hunch that current LLMs might make it easier to launch a brand new programming language, provided you can describe it in a few thousand tokens and ship it with a compiler and linter that coding agents can use […]

New TIL: Using Codex CLI with gpt-oss:120b on an NVIDIA DGX Spark via Tailscale https://til.simonwillison.net/llms/codex-spark-gpt-oss

Notes on Kimi K2 Thinking, the huge new open weights (but not open source, it's under a "modified MIT license") model from Moonshot AI https://simonwillison.net/2025/Nov/6/kimi-k2-thinking/

RE: https://mastodon.social/@Sarahp/115504173496002298 Glad to see this become a universal feature as opposed to one that was limited to specific clients

Made a new video demonstrating my process for upgrading a Datasette plugin using uv and an OpenAI Codex bash one-liner https://www.youtube.com/watch?v=qy4ci7AoF9Y Here are detailed notes to accompany the video on my blog: https://simonwillison.net/2025/Nov/6/upgrading-datasette-plugins/

I've been getting a lot of value using coding agents for code research tasks recently - I have a dedicated simonw/research GitHub repo and I frequently have them run detailed experiments and write up the results. Here's how I'm doing that + some examples […]

Achievement unlocked: caused Hacker News to have a 150+ comment argument about a TikTok joke (while I was asleep) https://news.ycombinator.com/item?id=45820872

Datasette 1.0a20 is out, featuring an entirely new SQL-powered permissions system. This is by far the most ambitious project I've attempted with the help of coding agents (Claude Code and Codex CLI in this case) - here are detailed notes on how it works and what I learned along the way […]

I wrote up some notes on two new papers on prompt injection: Agents Rule of Two (from Meta AI) and The Attacker Moves Second (from Anthropic + OpenAI = DeepMind + others) https://simonwillison.net/2025/Nov/2/new-prompt-injection-papers/

Just sent out the October edition of my sponsors-only monthly newsletter - you can pay me $10/month to send you less! Here's the table of contents https://simonwillison.net/2025/Nov/1/sponsors-only-newsletter/

My notes on CoreWeave's acquistion of Marimo - this year they also snapped up Weights & Biases, OpenPipe and Mammoth AI https://simonwillison.net/2025/Oct/31/coreweave-acquires-marimo/

MiniMax M2 is the new "most intelligent" open weights model (according to Artificial Analysis) - the MIT licensed weights are just 230GB and it appears comparable to Sonnet 4, while priced closer to Gemini 2.5 Flash. Notes here, including a new LLM plugin […]

Notes on Cursor 2.0 and a pelican drawn by their brand new Composer-1 coding model, which they describe as "4x faster than similarly intelligent models" https://simonwillison.net/2025/Oct/29/cursor-composer/

The GitHub Universe badge this year is a full Raspberry Pi with a color screen and WiFi! I had a ton of fun hacking around with it yesterday, here are detailed notes on what I've built so far https://simonwillison.net/2025/Oct/28/github-universe-badge/

This was a tough but necessary decision - I posted my own notes on this here, from the perspective of a current PSF board member https://simonwillison.net/2025/Oct/27/psf-withdrawn-proposal/ https://fosstodon.org/@ThePSF/115446659188615376

It's neat how if you ask Claude Code questions about itself it can answer them, because it knows how to fetch a Markdown index of its own online documentation and then navigate to the right place I wish more LLM tools would implement the same pattern! […]

Geoffrey Litt just proposed a new analogy for working with AI coding tools that I really like: you are the surgeon, staying in command and doing the most challenging work - the AI tools are your support team and surgical assistants https://simonwillison.net/2025/Oct/24/geoffrey-litt/

I recorded a ten minute video showing my vibe-coding process for building a tool for sharing formatted terminal sessions via copy and paste using the new Claude Code for web - now available on YouTube here https://www.youtube.com/watch?v=GQvMLLrFPVI Here are detailed notes (including the full […]

OpenAI's CISO Dane Stuckey posted an essay (on Twitter) about how their new ChatGPT Atlas browser attempts to deal with the risk of prompt injection attacks, I ended up writing a point-by-point commentary on my blog: https://simonwillison.net/2025/Oct/22/openai-ciso-on-atlas/

I gave a talk last night about "Living dangerously with Claude", on the joys and perils of --dangerously-skip-permissions and how critical it is that we run coding agents in a sandbox so that we can unlock their full potential https://simonwillison.net/2025/Oct/22/living-dangerously-with-claude/

Just for fun, I had Claude Code figure out how to run the ~2001-era Perl and C SLOCCount program in WebAssembly in the browser, complete with a UI for counting source code lines from pasted text, a GitHub repository or a zip file […] [Original post on fedi.simonwillison.net]

Wrote up my first impressions of ChatGPT Atlas, OpenAI's new browser - I remain unconvinced by the entire category of "browser agents", the security and privacy challenges still feel insurmountable to me https://simonwillison.net/2025/Oct/21/introducing-chatgpt-atlas/

My notes on Claude Code for web, Anthropic's new asynchronous coding agent - I had preview access over the weekend, it's effectively a sandboxed instance of "claude --dangerously-skip-permissions" running in Anthropic's container https://simonwillison.net/2025/Oct/20/claude-code-for-web/

DeepSeek released a new OCR model - I got it working on an NVIDIA Spark (CUDA + ARM64) by starting a Docker container, running Claude Code as root with "claude --dangerously-skip-permissions" and telling it to figure it out Took 4 prompts and 40 minutes […]

New TIL: I spent some time exploring OpenAI's o4-mini-deep-research model via their Responses API and wrote up some notes on what I learned, including vibe-coding a few tools to help me evaluate the results https://til.simonwillison.net/llms/o4-mini-deep-research

Claude Skills are awesome, maybe a bigger deal than MCP https://simonwillison.net/2025/Oct/16/claude-skills/

My notes (and a pelican) for the new Claude Haiku 4.5 model, released by Anthropic this morning https://simonwillison.net/2025/Oct/15/claude-haiku-45/

Does widespread browser implementation of the Sec-Fetch-Site HTTP header mean we can protect against CSRF attacks without needing those hidden form tokens? It looks like the answer may be a cautious "yes"! https://simonwillison.net/2025/Oct/15/csrf-in-go/

NVIDIA sent me preview hardware of their new DGX Spark 128GB ARM64 4TB "AI supercomputer" - it's a very neat little device, here are my notes so far https://simonwillison.net/2025/Oct/14/nvidia-dgx-spark/

nanochat by Andrej Karpathy is neat - 8,000 lines of code (mostly Python, a tiny bit of Rust) that can train an LLM on $100 of rented cloud compute which can then be served with a web chat UI on a much smaller machine https://simonwillison.net/2025/Oct/13/nanochat/

I'm beginning to suspect that a key skill in working effectively with coding agents is developing an intuition for when you don't need to closely review every line of code they produce. This feels deeply uncomfortable!

TIL Claude's new code interpreter mode has a /mnt/skills/public/ folder full of prompt instructions and Python utilities for creating and manipulating pdf, docx, pptx, xlsx files - and you can ask Claude for a copy and learn a TON about working with those formats […]

This is a wildly creative set of customizations for Claude Code, using the new plugin system they just released. There are SO many fascinating ideas in this! Strongly recommend reading it and then spending some time exploring the accompanying repo: https://github.com/obra/superpowers […]

I figured out a uv recipe for running the tests for the current project (any directory with a pyproject.toml or setup.py) using any version of Python: uv run --python 3.14 --isolated --with-editable '.[test]' pytest I've wrapped it as a script, uv-test, so I can do this instead: uv-test -p 3 […]

This isn't exactly a surprise, but here's a note that Claude 4.5 Sonnet in Claude Code can write a full working Datasette plugin from a single prompt now https://simonwillison.net/2025/Oct/8/claude-datasette-plugins/

Python 3.14 is out today! Here are my notes on the new release: https://simonwillison.net/2025/Oct/8/python-314/ If you're an open source library maintainer who supports all current Python releases this also means you can drop 3.9 support now and start depending on features from 3.10, like […]

I just tried the official demo for the new Gemini 2.5 Computer Use model and it started by navigating to Google, solving Google's own CAPTCHA and then running a search! https://simonwillison.net/2025/Oct/7/gemini-25-computer-use-captchas/

"brain coding" is an excellent term https://social.linux.pizza/@domm/115327839682387145

Vibe coding is irresponsibly building software through dice rolls, not caring what code is produced What about when engineers at the top of their game use AI tools responsibly to accelerate their work? I propose "vibe engineering"! (with my tongue only partially in my cheek) […]

I got the new GPT-5 pro API model to "Generate me an SVG of a pelican riding a bicycle". This pelican took 6m8s to generate and cost me $1.10! https://simonwillison.net/2025/Oct/6/gpt-5-pro/

Decided to live blog this morning's OpenAI DevDay announcements, since I'm in the audience https://simonwillison.net/2025/Oct/6/openai-devday-live-blog/

I've become one of those people who run multiple coding agents at once - here are my notes on embracing the parallel coding agent lifestyle https://simonwillison.net/2025/Oct/5/parallel-coding-agents/

If you've been trying to figure out DSPy - the automatic prompt optimization system - this talk by @dbreunig is the clearest explanation I've seen yet, with a very useful real-world case study https://www.youtube.com/watch?v=I9ZtkgYZnOw My notes here […]

Joshua Rogers, using AI tooling responsibly and professionally, reported 22+ genuine issues in curl that are now being addressed This is especially notable because curl has had huge problems with floods of garbage slop AI "security issues" in the past that were nothing of the sort - my notes […]