Every post of theirs in the library
Loading their posts
Simon Willison
Writes about AI, Engineering and Revenue. 75% of posts open with an observation. A typical post reaches 480 likes.
- Posts
- 52
- Likes
- 39K
- Typical
- 480
- Best
- 4.9K
What they write about
How they open
Every post of theirs in the library
| Post | Topic | Likes | Replies | Reposts | Score | Posted | Actions |
|---|---|---|---|---|---|---|---|
| @simonwHere's my enormous round-up of everything we learned about LLMs in 2025 - the third in my … | AI | 4.9K | 101 | — | 58 | ||
| @simonwThis one is pretty nasty - it tricks Antigravity into stealing AWS credentials from a .env… | AI | 2.2K | 49 | — | 51 | ||
| @simonwOpenAI aren't talking about it yet, but it turns out they've adopted Anthropic's brilliant… | AI | 2.1K | 76 | — | 51 | ||
| @simonwA few notes on Anthropic's new Claude Fable 5.1 - with Max thinking level I got the best S… | AI | 2K | 124 | — | 59 | ||
| @simonwIf you try nothing else today, give the demo at https://t.co/kUyz0fxIfE a go - it lets you… | AI | 2K | 53 | — | 47 | ||
| @simonwThis is so confusing. Did Anthropic really just drop Claude Code from their $20/month plan… | AI | 1.5K | 184 | — | 50 | ||
| @simonwThe conclusions here feel wrong to me. The two lessons I see are: 1. Don't run agents anyw… | AI | 1.4K | 166 | — | 50 | ||
| @simonwFive new Mistral models today, but the one I'm most excited about is this tiny 3B model th… | AI | 1.3K | 21 | — | 47 | ||
| @simonwTucked away in this article is an appeal to the AI skeptics to PLEASE stop writing off sto… | AI | 1.2K | 95 | — | 47 | ||
| @simonwThis is notable: Opus 4.5 is ~60% more expensive than Sonnet ($25/million output compared … | AI | 1.1K | 43 | — | 46 | ||
| @simonwThis is so wild... the leaked Opus soul document has now been confirmed! I wrote some init… | AI | 1.1K | 23 | — | 46 | ||
| @simonwThese papers are fascinating, but my favorite thing about them is they aren't PDFs! They'r… | Science | 1.1K | 20 | — | 41 | ||
| @simonwSystem prompt: "If the person is unnecessarily rude, mean, or insulting to Claude, Claude … | AI | 990 | 49 | — | 45 | ||
| @simonwThe "introductory pricing" for the 3.7 Flash model is really weird It's scheduled to doubl… | AI | 986 | 110 | — | 50 | ||
| @simonwDeepSeek released a new OCR model - I got it working on an NVIDIA Spark (CUDA + ARM64) by … | AI | 934 | 26 | — | 44 | ||
| @simonwI was disappointed by OpenAI's thin initial post on the sycophancy behavior bug, but it tu… | AI | 723 | 13 | — | 38 | ||
| @simonwI used Claude Code to reverse-engineer the Claude macOS Electron app and had Cowork dig ar… | AI | 715 | 19 | — | 42 | ||
| @simonwFascinating new thing from Anthropic: you can now build custom programs that speak a speci… | AI | 688 | 16 | — | 38 | ||
| @simonwMIcrosoft's Phi-3 really is an astonishingly good model - MIT licensed and small enough to… | AI | 665 | 12 | — | 38 | ||
| @simonwSome of the backlash I'm seeing against Apple AI is reinforcing my opinion: The hardest pr… | AI | 622 | 71 | — | 39 | ||
| @simonwThis is a legitimately fascinating leaked document, it appears to be the onboarding guide … | SaaS | 609 | 12 | — | 37 | ||
| @simonwI spent some time yesterday exploring the new Amazon Nova LLM family, and I'm really impre… | AI | 606 | 26 | — | 38 | ||
| @simonwI think the most notable things about o3-mini may be the price - less than half that of GP… | AI | 592 | 28 | — | 37 | ||
| @simonwNotes on today's DeepSeek v3 0324 model - a 641 GB MIT licensed monster, but you can run i… | AI | 584 | 17 | — | 37 | ||
| @simonwThis is detailed, fascinating and answers all sorts of open questions I'd love to know mor… | AI | 519 | 38 | — | 40 | ||
| @simonwWe have a few of these browser agent products now - Perplexity Comet and Brave Leo and Cla… | AI | 490 | 163 | — | 44 | ||
| @simonwWith Atlas being retired in favor of the browser embedded in the ChatGPT app I wonder if t… | AI | 470 | 62 | — | 40 | ||
| @simonwDeepSeek released a whole family of inference-scaling / "reasoning" models today, includin… | AI | 463 | 12 | — | 35 | ||
| @simonwMy hunch for now is that this was an ill-considered test which they didn't anticipate woul… | AI | 443 | 19 | — | 38 | ||
| @simonwI really wish @OpenAI would give the new image generation feature in GPT-4o a usable name … | AI | 438 | 97 | — | 38 | ||
| @simonwAnyone know if ChatGPT Codex (in the new ChatGPT desktop app) is a strict superset of Chat… | Engineering | 411 | 57 | — | 39 | ||
| @simonwToday in definitions of "agent", we had: 1. "systems that independently accomplish tasks o… | AI | 383 | 28 | — | 34 | ||
| @simonwYou can try out the mysterious gpt2-chatbot at https://t.co/4ARf9G6RHb (select "Direct Cha… | AI | 380 | 17 | — | 34 | ||
| @simonwMIT licensed model, their self-reported benchmarks show it performing similar to Claude So… | AI | 360 | 18 | — | 36 | ||
| @simonwI've been throwing a bunch of coding tasks at the new o1 and it feels like it may be equiv… | AI | 345 | 26 | — | 34 | ||
| @simonwThis sounds a lot more exciting than it is - o1 didn't have the ability to do these things… | AI | 334 | 4 | — | 32 | ||
| @simonwPelicans for Gemma 4 E2B, E4B, 26B-A4B and 31B - the first three generated on my laptop vi… | Engineering | 328 | 26 | — | 36 | ||
| @simonwNotes (and a Pelican) on Claude Sonnet 5 - the new tokenizer makes it ~1.4x more expensive… | AI | 325 | 42 | — | 37 | ||
| @simonwUnder-reported details of the xAI/Anthropic Colossus data center deal: Anthropic get Colos… | AI | 300 | 20 | — | 35 | ||
| @simonwClaude 3.5 Haiku is out - two surprises: 1. It's priced differently from Claude 3 Haiku. 3… | AI | 283 | 14 | — | 32 | ||
| @simonw128,000 token length, weights available under a non-commercial license or a commercial API… | AI | 279 | 3 | — | 31 | ||
| @simonwI was a guest on @lennysan's podcast! We talked about agentic engineering and all sorts of… | AI | 270 | 24 | — | 35 | ||
| @simonwMy notes on this morning's OpenAI release of the new GPT-4o model - not a huge leap in "in… | AI | 231 | 7 | — | 30 | ||
| @simonwI keep seeing Microsoft's Recall described as an "AI" feature, anyone know what makes it "… | AI | 212 | 32 | — | 31 | ||
| @simonw"a new 70B model that delivers the performance of our 405B model" is exciting because I mi… | AI | 211 | 7 | — | 29 | ||
| @simonwMeta just dropped Llama 4 on a weekend! Two new open weight models (Scout and Maverick) an… | AI | 193 | 5 | — | 28 | ||
| @simonwI really hope we get details from @OpenAI on the task that as specifies to their rogue age… | AI | 183 | 28 | — | 33 | ||
| @simonwqwq is a brand new openly licensed LLM from Alibaba Cloud's Qwen team in China. It's an op… | AI | 180 | 7 | — | 28 | ||
| @simonwMy notes on Gemini 3.5 Flash - 3x the price of Gemini 3 Flash but Google are planning to u… | Revenue | 170 | 23 | — | 32 | ||
| @simonwGemini 2.0 Flash is now available via their API (multi-modal input and text output only fo… | AI | 168 | 6 | — | 27 | ||
| @simonwOne of the most notable trends from 2024 was the total collapse in terms of LLM pricing - … | AI | 162 | 4 | — | 27 | ||
| @simonwFascinating insight from Anthropic here concerning how Claude is actually being used, via … | AI | 160 | 1 | — | 26 |