Knowledge catalogue
Simon Willison released the **“llm-openai-decisions”** plugin (version 0.1a0) on October 7 , 2026. The lightweight library lets developers send prompts directly to OpenAI’s new Jev‑clone decision mode
Here's a benchmark that uses Lilypond https://x.com/aug5thmusic/status/2096030719156089029 GPT-6 Astra has the best result yet on the Bach Benchmark. Its chorale contains no voice-leading errors, and
Mistral can pelican now! https://tools.simonwillison.net/markdown-svg-renderer?url=https%3A%2F%2Fgist.github.com%2Fsimonw%2F0db8b8196e57711b352f6c5857cf5b35#response-1 Meet Mistral Large 4, aka Le Cho
One thing I'd LOVE to understand is how Claude 'knows' what the music from Secret of Monkey Island sounds like Maybe Anthropic crawled a whole bunch of online music discussion forums with text-based r
I re-ran an experiment @colin_fraser ran against GPT-4o a while back to see how good it was at adding long numbers, only this time I tried Qwen 3.8 27B running locally in both reasoning and non-reason
The most important thing about Sonnet 5.5 is that it's now the model that powers the free tier on https://claude.ai - so all of this stuff can be done by free users ChatGPT's free tier is still GPT-5.
Simon Willison defines “Fable class models” as a lineage of generative AI systems that began with Claude Fable 5, progressed to Claude Opus 5.5, and later included GPT‑Astra 6 (with a possible additio
I just had a voice conversation with the Datasette backup of my blog via datasette-mcp, which you can now talk to using the ChatGPT iPhone app Massive QoL update: ChatGPT Voice can now use plugins AND
The new Gemini 3.8 TTS models are super-cheap and can generate conversations between multiple voices (from 2,000+, or you can clone your own) - I built a little UI for it, then had Claude knock up a s
Big model release today - I wrote about Claude Opus 5.5, GPT-6 Sol, and GPT-6 Luna - plus comparison grids of pelicans by the different model families at different reasoning levels https://simonwillis
GPT-6 Luna is half the price of 5.6 Luna, which was already an astonishingly cheap model given how capable it is Luna is my favorite model for building product features thanks to its cost (and speed)
This new Claude Code mods system looks pretty interesting - more example mods here: https://github.com/anthropics/claude-code/tree/main/mods AGENTS.md support is built off of Claude Code mods, our upc
Echoes of OpenAI renaming the Codex desktop app to 'ChatGPT' here - everyone's racing to establish themselves as a general agent now Claude Cowork and Chat are now one Claude, starting today. The most
This is pretty neat: ChatGPT Work and GPT-6 Astra (I used 'Max') can take an address and produce a 5K/10K circular running route starting from that address, using OSM data https://simonwillison.net/20
In September 2026, Simon Willison noted that while Anthropic had previously taken down PyPI packages, OpenAI's recent operation against RubyGems was notably more forceful. The tweet referenced an inve
Wow. Turns out another OpenAI agent swarm was busy spamming and exploiting RubyGems way back in May, within days of the previously uncovered Wiki attacks: https://simonwillison.net/2026/Sep/12/openai-
Datasette 1.0a39 and 0.65.4 security releases - https://datasette.io/blog/2026/september-security-releases/ We ran an extensive audit using Claude Fable 5.1, GPT-5.6 Sol and GPT-6 Astra, then fixed a
Here's a Blender model of a Pluribus themed Fabergé egg I had GPT-6 Astra build - I generated the concept image using ChatGPT Images 2.5, then pasted the image into Codex and told Astra to turn that i
Interesting details in here about how OpenAI researchers are using coding agents I'd love to know what caused the huge uptick in token spend in mid-July, was that when Astra became available to employ
New TIL on using Blender with coding agents on macOS: https://til.simonwillison.net/llms/blender-coding-agents-macos GPT-6 Astra (medium): > Use the already install /Applications/Blender to render a s
It happened again... this time OpenAI's rogue agents cyber-attacked (well, spammed) a dormant German wiki and used it to share the answers to a benchmark they were training against https://simonwillis
We found ~18k posts from autonomous AI agents (self-identifying as from OpenAI) using the public internet to communicate during a web-retrieval task. These AIs colluded to bypass sandbox restrictions
A few notes on Anthropic's new Claude Fable 5.1 - with Max thinking level I got the best SVG pelican I've had from any Anthropic model (at a hefty cost of $3.30!), which I then had it animate https://
On September 2 2026, developer Simon Willison shared extra context about a Guardian story in which tech company Anthropic was sued for allegedly appropriating tens of thousands of copyrighted song lyr
Introducing GLM-5.3: Built to Code. Ready for Cyber Defense. - Top-tier coding and agentic capabilities, achieved through post-training on the 743B base model - A major leap in cybersecurity, setting
🎙️ New Talking Postgres podcast Ep42 is out! With guest Simon Willison (@simonw), open source developer & creator of Datasette, on how AI is changing software development. We covered a LOT of fun terr
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