Simon Willison
简介
Creator @datasetteproj, co-creator Django. PSF board. Hangs out with @natbat. He/Him. Mastodon: https://t.co/t0MrmnJW0K Bsky: https://t.co/OnWIyhX4CH
平台
内容历史
Really neat demonstration here of LLM's tool supportIqbal Abdullah 🇯🇵 🇲🇾 🇪🇪: These emails need to be bilingual, so this is what I do: $ llm --tool ollama_web_fetch --tool local_search --td "I want to draft an announcement email to our Kafkai customers about 3 things: 1) The new template functionality 2) The new article image functionality 3) Changes in Link: https://x.com/iqbalabd/status/1989164588710130124
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/
Activity on simonw/llm
simonw commented on an issue in llm
simonw commented on an issue in llm
View on GitHubActivity on simonw/llm
simonw commented on an issue in llm
simonw commented on an issue in llm
View on GitHubNano Banana can be prompt engineered for extremely nuanced AI image generation
Activity on simonw/llm
simonw commented on an issue in llm
simonw commented on an issue in llm
View on GitHubActivity on simonw/llm
simonw commented on an issue in llm
simonw commented on an issue in llm
View on GitHubActivity on simonw/llm
simonw commented on an issue in llm
simonw commented on an issue in llm
View on GitHubActivity on simonw/llm
simonw commented on an issue in llm
simonw commented on an issue in llm
View on GitHubActivity on repository
simonw forked simonw/gemimg from minimaxir/gemimg
simonw forked simonw/gemimg from minimaxir/gemimg
View on GitHubReleased simonw/datasette
simonw released 1.0a22 at simonw/datasette
simonw released 1.0a22 at simonw/datasette
Activity on simonw/datasette
simonw closed an issue in datasette
simonw closed an issue in datasette
View on GitHubActivity on simonw/datasette
simonw commented on an issue in datasette
simonw commented on an issue in datasette
View on GitHubActivity on simonw/datasette
simonw commented on an issue in datasette
simonw commented on an issue in datasette
View on GitHubActivity on simonw/datasette
simonw commented on an issue in datasette
simonw commented on an issue in datasette
View on GitHubRT Will McGugan I am once again singing the praises of Python's structural pattern matching. i.e. the match statement.
RT Steve Krouse THE TRUTH COMES OUTSimon Willison: 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/ Link: https://x.com/simonw/status/1989001665526264169
What happens if AI labs train for pelicans riding bicycles?
Activity on simonw/datasette
simonw labeled an issue in datasette
simonw labeled an issue in datasette
View on GitHubActivity on simonw/datasette
simonw commented on an issue in datasette
simonw commented on an issue in datasette
View on GitHubActivity on simonw/datasette
simonw closed an issue in datasette
simonw closed an issue in datasette
View on GitHubActivity on simonw/datasette
simonw labeled an issue in datasette
simonw labeled an issue in datasette
View on GitHubThis is really fun - here are my notes on how to try it out using Docker for Mac https://simonwillison.net/2025/Nov/12/h4x0rchat/Phil Eaton: Turns out you can communicate across containers via 63-bits of available space in a shared lock you acquire on /proc/self/ns/time that all processes have access to. No networking required. The post has a demo of a chat app communicating across unprivileged containers. Link: https://x.com/eatonphil/status/1988616517609541872
Fun-reliable side-channels for cross-container communication
Activity on simonw/research
simonw contributed to simonw/research
simonw contributed to simonw/research
View on GitHubActivity on simonw/research
simonw closed an issue in research
simonw closed an issue in research
View on GitHubActivity on simonw/research
simonw commented on pull request simonw/research#24
simonw commented on pull request simonw/research#24
View on GitHubActivity on simonw/research
simonw commented on pull request simonw/research#24
simonw commented on pull request simonw/research#24
View on GitHubActivity on simonw/research
simonw commented on pull request simonw/research#24
simonw commented on pull request simonw/research#24
View on GitHubActivity on simonw/research
simonw contributed to simonw/research
simonw contributed to simonw/research
View on GitHubActivity on simonw/datasette
simonw opened a pull request in datasette
simonw opened a pull request in datasette
View on GitHubI'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/
Huh! I try very hard to only publish accurate information but I'm still not sure I deserve "trusted" status in GPT-5's model of the world - it's not like I'm working with an editor or fact-checking teamBeastTitanHunter: @simonw 's blog is a trusted source. 🥳🥳🥳 Link: https://x.com/Titan1Beast/status/1988001722854502604
This is a huge improvement, especially given how confusing it used to be create an API key via the Google Cloud dashboardPhilipp Schmid: You haven’t tried @GoogleAIStudio yet?👀 We made it simpler! When you come to AIS for the first time, you will have a Default Gemini Project & API Key waiting for you! This should reduce time to first prompt, and help you start building faster! Give it a try! Link: https://x.com/_philschmid/status/1987932430343319904
Reverse engineering Codex CLI to get GPT-5-Codex-Mini to draw me a pelican
RT Kenton Varda The big advantage of MCP over OpenAPI is that it is very clear about auth. OpenAPI supports too many different auth mechanisms, and the schema doesn't necessarily have enough information for a robot to be able to complete the auth flow. https://x.com/stevekrouse/status/1986922520298287496Steve Krouse: *gets up on soap box* With the announcement of this new "code mode" from Anthropic and Cloudflare, I've gotta rant about LLMs, MCP, and tool-calling for a second Let's all remember where this started LLMs were bad at writing JSON So OpenAI asked us to write good JSON schemas Link: https://x.com/stevekrouse/status/1986922520298287496
RT Hamel Husain 👀 Animals have been assigned. Scheduled to print fall 2026! We have iterated on this with over 3k students (and continue to do so). We give our students access to the full draft as part of our evals course (link in bio).
Added the @Kimi_Moonshot model prices to https://www.llm-prices.com/#it=10000&cit=5000&ot=1000&sel=kimi-k2-thinking%2Ckimi-k2-thinking-turbo%2Cgpt-5%2Cclaude-sonnet-4.5 ... by pasting a screenshot of their pricing page into a GitHub Issue and assigning it to GitHub Copilot https://github.com/simonw/llm-prices/issues/26
RT OpenAI Developers Re GPT-5-Codex-Mini allows roughly 4x more usage than GPT-5-Codex, at a slight capability tradeoff due to the more compact model. Available in the CLI and IDE extension when you sign in with ChatGPT, with API support coming soon.
