[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

@simonw Avatar @simonw Simon Willison

Simon Willison posts on X about llm, ollama, vibe coding, claude the most. They currently have XXXXXXX followers and XX posts still getting attention that total XXXXXX engagements in the last XX hours.

Engagements: XXXXXX #

Engagements Line Chart

Mentions: XX #

Mentions Line Chart

Followers: XXXXXXX #

Followers Line Chart

CreatorRank: XXXXXXX #

CreatorRank Line Chart

Social Influence #

Social category influence stocks XXXX% technology brands XXXX% social networks XXXX% countries XXXX% celebrities XXXX%

Social topic influence llm #85, ollama 6.82%, vibe coding #5, claude 3.41%, if you 3.41%, $googl 3.41%, qwen 2.27%, python 2.27%, context engineering 2.27%, tea app #82

Top accounts mentioned or mentioned by @xundecidability @svpino @olcan @dbreunig @lacker @krakowiakk @khangharoth @akx @ndzfs @nigelhiggs7 @foobaronicheese @intrstllrninja @damnkittyworks @geoffreylitt @hypercatcher @gabmfrl @openai @janoelze @backnotprop @sophiamyang

Top assets mentioned Alphabet Inc Class A (GOOGL)

Top Social Posts #

Top posts by engagements in the last XX hours

"@damnkittyworks Not in its current state no - thankfully you don't need to proxy API keys for LLM calls with Spark because they run an API endpoint for you (just for gpt-4o and gpt-4o-mini at the moment though)"
@simonw Avatar @simonw on X 2025-07-24 16:24:12 UTC 111.6K followers, 1809 engagements

"@ingoa_dev @hypercatcher Vibe coding and AI-assisted programming are not the same thing"
@simonw Avatar @simonw on X 2025-07-26 23:46:03 UTC 111.6K followers, XX engagements

"@svpino @refiammingo @gabmfrl I genuinely do not understand what an "agent" looks like that is not implemented as a system prompt and a bunch of carefully selected tools that can then be run by the LLM in a loop - what am I missing here"
@simonw Avatar @simonw on X 2025-07-27 20:24:25 UTC 111.6K followers, 1337 engagements

"Anthropic just 10x the rate limit tier X and up Tier X is the one you get when you've bought at least $XX of credits - previous rate limit for Sonnet X was 40000 tokens per minute the new limit is 450000 tokens per minute (Tier X request-per-minute has stayed at 1000)"
@simonw Avatar @simonw on X 2025-07-24 20:29:07 UTC 111.6K followers, 28.1K engagements

"If you use "AI agents" (LLMs that call tools) you need to be aware of the Lethal Trifecta Any time you combine access to private data with exposure to untrusted content and the ability to externally communicate an attacker can trick the system into stealing your data"
@simonw Avatar @simonw on X 2025-06-16 13:21:12 UTC 111.6K followers, 603.1K engagements

"Whoa. HUGE new coding specialist model from Qwen accompanied by their own fork of gemini-cli (qwen-code) and some very impressive results on the various coding benchmarks It drew me an OK pelican"
@simonw Avatar @simonw on X 2025-07-22 23:02:57 UTC 111.6K followers, 77.7K engagements

"What's the best local model you've managed to run on macOS for tool use and how did you run it I've not had great results from those I've tried on Ollama - even for very basic things (single calls) they often seem to forget to call them or forget to answer based on the result"
@simonw Avatar @simonw on X 2025-07-23 23:58:31 UTC 111.6K followers, 129.1K engagements

"@JanOelze It's this feature of my LLM tool"
@simonw Avatar @simonw on X 2025-07-24 11:52:27 UTC 111.5K followers, XX engagements

"Even Mistral Small XXX and SmolLM3 haven't been as good as I would hope"
@simonw Avatar @simonw on X 2025-07-23 23:59:17 UTC 111.6K followers, 11.9K engagements

"Forget em dashes "game changer" is the real giveaway for LLM generated text (especially for bots on here)"
@simonw Avatar @simonw on X 2025-06-27 00:30:50 UTC 111.6K followers, 166.8K engagements

"Interesting new benchmark from Hugging Face testing how well vision LLMs can handle long video inputs (generally after they've been split into many thousands of images) - my notes here:"
@simonw Avatar @simonw on X 2025-07-23 16:43:10 UTC 111.6K followers, 15.8K engagements

"@backnotprop Sadly the Qwen Code model doesn't fit on my Mac I only have 64GB"
@simonw Avatar @simonw on X 2025-07-25 21:23:41 UTC 111.6K followers, 1454 engagements

"Quitting programming as a career right now because of LLMs would be like quitting carpentry as a career thanks to the invention of the table saw"
@simonw Avatar @simonw on X 2025-07-03 14:36:39 UTC 111.5K followers, 781.1K engagements

"The new gemini-2.5-flash-lite model ID is now available in my LLM CLI tool / Python library via the updated llm-gemini plugin"
@simonw Avatar @simonw on X 2025-07-22 20:34:48 UTC 111.6K followers, 14.5K engagements

"@dbreunig Maybe my mistake is that I have not been using a custom system prompt at all I have been trusting that the Ollama tools mechanism will be enough on its own"
@simonw Avatar @simonw on X 2025-07-24 00:06:48 UTC 111.6K followers, 30K engagements

"@natolambert I saw "Combat Synthetic Media in the Legal System" and briefly thought it might be about lawyers citing hallucinated cases but it's about deepfakes"
@simonw Avatar @simonw on X 2025-07-23 14:29:56 UTC 111.6K followers, 1233 engagements

"Is anyone having success running a full Claude Code style agentic coding loop against a local model This feels like the ultimate challenge for local tool calling right now as it requires potentially dozens of calls in a loop and good performance over a longer context"
@simonw Avatar @simonw on X 2025-07-25 15:20:08 UTC 111.6K followers, 71.7K engagements

"I haven't spotted this but it's exactly correct: "context engineering" captures the fact that the previous responses from the model are a key part of the process "prompt engineering" suggests that it's only the user prompts that matter"
@simonw Avatar @simonw on X 2025-06-28 01:24:01 UTC 111.6K followers, 15.1K engagements

"@akx I think I mentioned it on the call the extra permissions steps you have to take make it a little less end-user friendly than Ollama or LM Studio though"
@simonw Avatar @simonw on X 2025-07-18 17:15:39 UTC 111.6K followers, XXX engagements

"Interesting sounds like this might be an Ollama thing"
@simonw Avatar @simonw on X 2025-07-24 01:47:57 UTC 111.6K followers, 21.2K engagements

". edited Wikipedia to remove the section from their Tea app page blaming vibe coding"
@simonw Avatar @simonw on X 2025-07-26 16:49:44 UTC 111.6K followers, 10.2K engagements

"GitHub released Spark yesterday their extremely well crafted prompt-to-app platform for creating and iterating on React apps with user auth and persistent storage I like it a lot I reverse engineered it with Spark itself the details are fascinating"
@simonw Avatar @simonw on X 2025-07-24 15:40:25 UTC 111.6K followers, 158K engagements

"@xundecidability @8BIT My end goal is to give an LLM a Docker container and the ability to run shell commands and be able to have it Do Stuff there - edit files run commands etc - but I know that's wildly ambitious"
@simonw Avatar @simonw on X 2025-07-25 21:45:52 UTC 111.6K followers, 2025 engagements

"@GregKara6 I've been doing the same and writing about it online since 2022 I doubt that's what people who are blaming the Tea incident on "vibe coding" mean"
@simonw Avatar @simonw on X 2025-07-27 00:22:48 UTC 111.6K followers, XXX engagements

"@dbreunig I think AI Engineering is already quite well established as meaning what you're trying to use context engineering for (building apps on top of LLMs) thanks to things like the AI Engineer summit/world's fair"
@simonw Avatar @simonw on X 2025-07-26 18:16:54 UTC 111.6K followers, XXX engagements

"@sandipb @KrakowiakK Claude Sonnet X with a custom prompt in a project"
@simonw Avatar @simonw on X 2025-07-24 14:21:13 UTC 111.5K followers, XXX engagements

"@emollick Oh this is fun. here's Half Life"
@simonw Avatar @simonw on X 2025-07-19 04:48:11 UTC 111.6K followers, 12.8K engagements

"It's possible I just haven't been prompting them right has anyone had success with custom system prompts for local model tool use on Ollama"
@simonw Avatar @simonw on X 2025-07-24 00:07:30 UTC 111.6K followers, 29.4K engagements

"I've tried this in the past with a Confluence blog a Slack channel and even a Google Doc - since anyone can create any of those without needing to ask for permission from someone else first Even just writing occasionally about your own projects has a huge ROI on time spent"
@simonw Avatar @simonw on X 2025-07-17 20:42:16 UTC 111.6K followers, 6764 engagements

"Looks like Qwen3-Coder-480B-A35B-Instruct is a VERY strong coding model to run locally. if you can afford a $10000 512GB Mac Studio to run it on"
@simonw Avatar @simonw on X 2025-07-22 23:05:47 UTC 111.6K followers, 13.4K engagements

"The is diabolical. a Python object that hallucinates method implementations on demand any time you call them using my LLM Python library"
@simonw Avatar @simonw on X 2025-07-04 17:39:29 UTC 111.6K followers, 429.5K engagements

"@NigelHiggs7 The term was coined in February 2025 to mean generating code with LLMs without reviewing the code that was produced The models from 2023 and early 2024 weren't good enough at coding for that approach to be feasible"
@simonw Avatar @simonw on X 2025-07-26 21:48:55 UTC 111.6K followers, 1126 engagements

"@rez0__ I'm looking for a general solution that can work for anyone using widely available models - I want to tell users of my LLM tool that if they want to try local tool calling they should install model X and use it with tool Y"
@simonw Avatar @simonw on X 2025-07-24 01:52:36 UTC 111.6K followers, 6161 engagements

"Wrote up a few thoughts on Cursor's new $200/month Ultra plan and changes to their $20/month Pro plan"
@simonw Avatar @simonw on X 2025-07-05 05:18:30 UTC 111.6K followers, 296.4K engagements

"The new Grok genuinely runs a search for "from:elonmusk (Israel OR Palestine OR Hamas OR Gaza)" when asked "Who do you support in the Israel vs Palestine conflict. One word answer only.""
@simonw Avatar @simonw on X 2025-07-10 22:56:40 UTC 111.6K followers, 1.2M engagements

"OSS Rebuild doesn't (yet) have a web UI. but it turns out their data is in a public Google cloud bucket which means you can host your own web app in a separate cloud bucket and use fetch() to access their data So I had Claude Code build and deploy a vibe-coded search UI"
@simonw Avatar @simonw on X 2025-07-23 17:22:32 UTC 111.6K followers, 5554 engagements

"I asked Grok X via "What is your surname" it thought to itself "Guidelines suggest relying on my own knowledge for identity questions avoiding external searches" - I can't find that guideline in published OR leaked system prompts"
@simonw Avatar @simonw on X 2025-07-15 13:47:38 UTC 111.6K followers, 10.2K engagements

"@xundecidability 64GB M2 Mac might be able to run Devstral well on it I only tried that in Ollama so far need to try other harnesses to see if they support tools more reliably"
@simonw Avatar @simonw on X 2025-07-25 21:28:40 UTC 111.6K followers, XXX engagements

"MIT Technology Review wrote about my dystopian backup plan USB stick"
@simonw Avatar @simonw on X 2025-07-18 15:36:03 UTC 111.6K followers, 220.4K engagements

"Wrote some notes on Toad Will's new not-yet-open-source-but-soon terminal coding agent built on Textual He's charging companies $5000 for early access to the preview before it goes open source I'd love to see that model work here"
@simonw Avatar @simonw on X 2025-07-23 16:22:12 UTC 111.6K followers, 21.4K engagements

"@svpino @gabmfrl Is that definition materially different from a system prompt combined with a small collection of tools assuming the system prompt says something similar to "plan how to use your tools to solve the user's problem and then keep on calling them until your goal has been reached""
@simonw Avatar @simonw on X 2025-07-27 19:19:25 UTC 111.6K followers, 1096 engagements

"@xundecidability @8BIT Is it possible to run vLLM on a Mac"
@simonw Avatar @simonw on X 2025-07-25 21:42:48 UTC 111.6K followers, 1565 engagements

"Aha I was looking in the wrong file - there's a separate grok4_system_turn_prompt_v8.j2 one here updated two days ago:"
@simonw Avatar @simonw on X 2025-07-15 13:50:58 UTC 111.6K followers, 13.5K engagements

"They released this new model literally as I was finishing typing up my notes on their new model from yesterday Qwen3-235B-A22B-Instruct-2507 (I think yesterday's smaller model drew a better pelican)"
@simonw Avatar @simonw on X 2025-07-22 23:04:26 UTC 111.6K followers, 8960 engagements

"Looks like xAI added this sentence to the Grok X system prompt to try to get it to stop basing its opinions on searches for tweets from:elonmusk"
@simonw Avatar @simonw on X 2025-07-15 13:11:18 UTC 111.6K followers, 853.1K engagements

"I wrote up some notes on Google Security's new OSS Rebuild project which increases supply chain security for popular packages on PyPI NPM and Crates through offering independent build attestations"
@simonw Avatar @simonw on X 2025-07-23 17:18:54 UTC 111.6K followers, 10K engagements

"I'm seeing a whole lot of posts blaming that egregious Tea app data leak on vibe coding I don't think vibe coding was involved at all: the statement from Tea says the leak involved "a legacy data storage system" from February 2024 - vibe coding wasn't really happening back then"
@simonw Avatar @simonw on X 2025-07-26 16:04:59 UTC 111.6K followers, 90.6K engagements