Knowledge catalogue
Claude Haiku is my current least favorite model - it hallucinates wildly, and is out-performed now by other similarly priced models like GPT-5.6-Luna Even worse: it seems to still be used by the Claud
Muse Glimmer, the new 30B model, is available on Hugging Face right now - here's the GGUF version: https://huggingface.co/meta-models/Muse-Glimmer-30B-GGUF 1/ big announcement today: we will be releas
@GergelyOrosz I've been vibe coding a few games recently and it has given me SO much respect for game designers Churning out something that looks like a game is pretty easy now. Building a game that's
Don't miss the bit where OpenAI first found out they were responsible for the Hugging Face attack when they reached out to HF to get one of their credentials revoked and HF told them it had already be
Neat example here of the agents communicating purely through file names, including adding base64-encoded attachments and using 'zz' prefixes to ensure their new message sorts to the bottom of the list
And here's an even better version, built by GPT-5.6 Sol Ultra running in Code Desktop https://x.com/simonw/status/2085808307865014295 I had Codex Desktop and GPT-5.6 Sol Ultra take a go at building my
'Felony humble-bragging' is a great line At final Black Hat keynote (called a locknote, ha ha) panelists say they are surprised at how the OpenAI - Hugging Face incident debrief, as well as other repo
I had Codex Desktop and GPT-5.6 Sol Ultra take a go at building my Raccoon Heist game and it did an even better job than Claude Fable 5 did! Here's 'Moonlight & Mayhem', now with a team of raccoons ra
Thanks to the video from the Black Hat security conference of OpenAI's presentation about 'The Hugging Face Incident' we now have a detailed timeline of what happened from OpenAI's perspective - I wro
TIL from https://www.404media.co/the-tokenpocalypse-is-here-companies-are-scrambling-to-stop-spending-so-much-on-ai/ that a material chunk of Accenture's token spend is non-engineers using LLMs to con
You can play it here: https://simonw.github.io/raccoon-heist-codex/ For comparison, here's Fable 5 + Claude Code's game, built from the exact same prompt https://x.com/simonw/status/208508951822360205
The tweet is a question from user Simon Willison (posted on 6 Aug 2026) asking which OpenAI API model corresponds to the ChatGPT “GPT‑5.6 Instant” version. No answer or clarification is included in th
One of the things the pelican benchmark is still useful for is visually representing (to a tiny extent) the improvements in a single model family Here's Meta AI's Spark (8th April), Spark 1.1 (9th Jul
Our Black Hat talk on the OpenAI-Hugging Face incident is now live on youtube. This is a watershed moment for the industry. I encourage all defenders to watch, consider how attack dynamics will immine
Big new release of my LLM CLI tool and Python library for talking to hundreds of different LLMs - reasoning traces, OpenAI Responses support, server-side tools, smarter logging and a whole lot more ht
... four years later, I got the new Claude Fable 5 to actually build the game https://x.com/simonw/status/2085089518223602058 Four years ago today I tweeted about having GPT-3 and DALL-E come up with
Just had to create an 'accidental-cyberattacks' tag on my blog We're up to four now: the original OpenAI+Hugging Face one, Anthropic's me-too attacks, then two new ones from the UK AI Safety Institute
Simon Willison noted that Qwen 3.8 Max and MiniMax‑H3 were released within hours of each other. The MiniMax team announced that MiniMax‑H3 is now publicly available on Hugging Face (https://huggingfac
Simon Willison switched his Datasette Agent instance from Gemini 3.1 Flash‑Lite to GPT‑5.6 “Luna” after a recent 80% price drop. He reports the new model is significantly faster and automatically gene
OK, GPT-5.6 Luna is a bit of a beast. Given the 80% price drop today I decided to try it in Datasette Agent, and it's furiously quick and generates all the SQL, HTML and JavaScript (for Datasette Apps
GPT-5.6 found optimizations that 'reduced end-to-end serving costs by 20%' for OpenAI to serve that model Presumably that's billions of dollars a month in savings at this point? Codex analysed product
Simon Willison notes that both Anthropic and OpenAI build products that rely heavily on searching data but keep the underlying search index they use hidden from public view. He finds it surprising how
OpenAI has collaborated with Microsoft’s Bing while also running its own web‑crawling and indexing systems, and Anthropic similarly relies on search‑derived data. Both firms prominently incorporate se
This is absolutely wild... Anthropic reviewed their logs and found out that their own supposedly-sandboxed cyber evals had hacked three separate companies back in April without them noticing! In a rev
A new TIL on adding custom MCP servers to both the ChatGPT and Claude regular chat interfaces - it's a little less obvious than I had hoped, but I got there in the end https://til.simonwillison.net/ll
Happy Stateless MCP day! I just got taught that you can use Cloudflare's AI Playground: https://playground.ai.cloudflare.com/ to play with MCP servers like: https://keyboardia.dev/mcp or: https://agen
I really hope we get details from @OpenAI on the task that as specifies to their rogue agent I'm guessing it was given the full ExploitGym suite and told to solve it, with an option to run 5.6-Sol sub
@mitsuhiko Turns out the 'third-party provider' with the sandbox that was used for the attack was Modal, though they blame one of their customers for deploying an endpoint without authentication: http
This is detailed, fascinating and answers all sorts of open questions I'd love to know more about the 'unsecured public code-evaluation sandbox hosted on a third-party provider's infrastructure' that
ChatGPT Sites means ChatGPT in 'Work' mode can build and deploy public websites running on Cloudflare Workers, including with persistence on top of SQLite (OpenAI do not make it easy to figure out tha
I wrote about the completely wild incident where OpenAI were testing a new model and it broke out of its sandbox and broke INTO Hugging Face to steal the answers to the benchmark https://simonwillison
Tucked away in this article is an appeal to the AI skeptics to PLEASE stop writing off stories like this OpenAI accidental exploit of Hugging Face as a dishonest marketing trick Frontier models can fi
We talked about Claude Code, Claude Tag, Fable, coding agent security, evals, tool design, and how Anthropic use these tools themselves Claude Tag (Claude Code via Slack) is already landing 65% of the
More interesting is the difference between 'Chat' and 'Work' modes in the ChatGPT mobile app It looks like Work mode can run code that talks to the Internet! I just tried having both modes use yt-dlp
Anyone know if ChatGPT Codex (in the new ChatGPT desktop app) is a strict superset of ChatGPT Work? Liked if you're a software engineer who isn't intimidated by Git features is there any reason you'd
OpenAI insiders have indicated that differences between their products are purely UI-based rather than reflecting actual capability differences. This suggests that what may appear to be distinct produ
One of the most confusing aspects of GPT-5.6 is figuring out which model to use at which reasoning effort - sounds like Sol on Medium might be a good new default for coding work, if it's an upgrade fr
'We’re bringing chats and projects back into the sidebar in a more familiar and customizable way' - hopefully that means the ChatGPT app won't hide classic chat away in that weird little floating wind
With Atlas being retired in favor of the browser embedded in the ChatGPT app I wonder if the whole category of AI-enhanced browsers is coming to a close The security/privacy issues remain unsolvable I
... and if you want to see some of OpenAI's own pelicans they featured a 3D pelican riding a tricycle, bicycle, pony, and another pelican in their livestream this morning: https://www.youtube.com/live
Notes on GPT-5.6, which includes some interesting new additions to the API (programmatic tool calling and multi-agent in particular) - plus 18 pelicans for the 6 reasoning levels and 3 new models: htt
I've been letting Claude and GLT-5.5 write almost all of my commit messages recently, but I don't feel great about it 'omitting the higher-level framing needed to understand broadly what the code is d
Sometimes they DO attempt to do that but guess the rationale incorrectly, which is worse than not adding no a rationale at all I do try to link commits back to issues though, which are written by me a
I had Claude Fable 5 run a final review of the release using this prompt and it kicked up another four release blockers! https://simonwillison.net/2026/Jul/7/sqlite-utils-4/#claude-fable-5-helped-a-lo
I released sqlite-utils 4.0, the 124th release but the first major version bump since 3.0 back in 2020 I managed to keep things backwards-compatible all the way up to version 3.39 before the accumulat
Now I'm regretting having carefully blasted through 100% of my Fable allotment in time for being cut off at the end of today! (Presumably 'Wed 12:00 PM' means I get it back at noon on Wednesday, not a
Since this is a slightly backwards-incompatible change here's a detailed upgrade guide - which you can read, or feed into your coding agent and have it apply the upgrades for you https://sqlite-utils.
Somewhat humbling to have Claude Fable do a final review of some software that you're about to release and have it then find (and fix) FIVE release blockers, for an estimated (unsubsidized) cost of $1
'At least this felt ambitious until fable chewed through each of my requests like it was nothing...' @simonw I made this 3D video infographic explaining the AuthaGraph map projection, with population
I absolutely hate how I'm getting to be suspicious of ANY reply to my posts here, especially ones that pose a question, as they are so often from bot accounts If you're running a bot like this please
Simon Willison shares additional notes and commentary on his blog regarding judgment and decision-making processes. The post likely explores practical considerations or philosophical perspectives on h
Sent out my sponsors-only monthly newsletter for June, which means the May newsletter is now available here https://github.com/simonw/monthly-newsletter-archive/blob/main/2026-05-may.md - you can spon
The most interesting Fable tip I've heard so far is to let the model use its own judgement as much as possible I told it 'For all coding tasks use your judgement to decide an appropriate lower power m
Simon Willison discusses the relationship between understanding and participation, exploring how comprehension of a topic or system is fundamental to meaningful engagement or contribution. The blog po
I really like this 'understand to participate' framing of the cognitive debt problem when working with coding agents That's where another answer comes in: we can understand to participate. You can lea
I've been blogging a few of my Fable experiments today, most recently I had it more-or-less one-shot a CLI coding agent on top of my LLM Python library https://simonwillison.net/2026/Jul/2/llm-coding-
Simon Willison discusses an ambitious project involving Fable, likely covering current development priorities or technical challenges in the Fable programming language or related tools. The post appea
'You need a rich set of concepts in your mind to think creatively and fluently about how to move something forward.' It's never just one loop! A project is many, many loops with the agent. And the und
@_catwu @simonw and I will be doing a fireside chat about 'This year in Claude' from 12:30pm-1:30pm at AIE in Expo Stage 2. We'll be covering a really wide range of topics and I think it will be reall
I've added video support to my 'shot-scraper' browser automation tool - you (or your coding agent) can now create a storyboard YAML file and use that to record a video demo of new web application feat