{
 "handle": "karpathy",
 "captured_at": "2026-10-12T15:35:00Z",
 "window": "2026-10-01 to 2026-10-12",
 "method": "Chrome extension DOM read of x.com (logged-in session) for timeline + status pages; cdn.syndication.twimg.com/tweet-result for per-ID verification",
 "note": "No posts between 2026-08-03 and 2026-10-01. Stats are as of capture time.",
 "posts": [
  {
   "id": "2109361546505966046",
   "url": "https://x.com/karpathy/status/2109361546505966046",
   "time": "2026-10-11T19:12:37Z",
   "kind": "quote-post (quotes his own 2026-04-09 post 2042334451611693415)",
   "stats": {
    "replies": 1643,
    "reposts": 6276,
    "likes": 35252,
    "bookmarks": 24591,
    "views": 4351495
   },
   "text": "Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less \"two groups of people speaking past each other\" and more a sharp funnel.\n\n- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.\n- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.\n- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.\n- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the \"archeology\" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.\n\nThe funnel is driven by a combination of factors:\n1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.\n2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.\n3. Access. Free-tier, paid-tier, internal.\n\nSo this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.",
   "about": "Quoting his own post from 9 April 2026",
   "thread_label": "The April post he quotes",
   "thread": "Judging by my tl there is a growing gap in understanding of AI capability.\n\nThe first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like \"should I drive or walk to the carwash\". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code.\n\nBut that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively \"peaky\" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along.\n\nSo that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of \"AI Psychosis\" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions.\n\nTLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned \"Advanced Voice Mode\" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are."
  },
  {
   "id": "2108727132918841657",
   "url": "https://x.com/karpathy/status/2108727132918841657",
   "time": "2026-10-10T01:11:41Z",
   "kind": "reply to @TheStalwart (Joe Weisenthal), who quoted @ankitkr0: 'every once in a while i think about jack clark, the bloomberg journalist who took a break to study AI for a few months and then ended up being one of cofounders of anthropic'. TheStalwart: 'Yeah I should have paid closer attention'",
   "stats": {
    "replies": 29,
    "reposts": 34,
    "likes": 1154,
    "bookmarks": 230,
    "views": 71527
   },
   "text": "I had some amount of journalistic inbound way back as a PhD student but the only person I consistently replied to was Jack because he actually put in effort, read academic papers and asked good technical questions and I have not really seen that before (or since).",
   "about": "Replying to @TheStalwart (Joe Weisenthal, Bloomberg)",
   "thread_label": "What he was replying to",
   "thread": "@ankitkr0: every once in a while i think about jack clark, the bloomberg journalist who took a break to study AI for a few months and then ended up being one of cofounders of anthropic\n\n@TheStalwart, reposting it: Yeah I should have paid closer attention"
  },
  {
   "id": "2106806571321966793",
   "url": "https://x.com/karpathy/status/2106806571321966793",
   "time": "2026-10-04T18:00:03Z",
   "kind": "reply to @omarsar0 (Elvis Saravia), who wrote about the Oct 2 'outputs' post: '50K likes on this!? Most of what Karpthy has shared is things we have been doing for over a year. No surprises, as many replies are highlighting. But I still think this raises an interesting discussion. Is there a universal interface/language that allows more seamless...'",
   "stats": {
    "likes": 2801
   },
   "text": "My mental model for what is happening is that 99%+ of people who are now paying attention have been onboarded to anything related to AI in <1 year. This is very confusing to the AI dinosaurs (anyone in AI for pre 2026. Or even, gasp, pre 2012).",
   "about": "Replying to @omarsar0 (Elvis Saravia)",
   "thread_label": "What he was replying to",
   "thread": "@omarsar0: 50K likes on this!? Most of what Karpthy has shared is things we have been doing for over a year. No surprises, as many replies are highlighting. But I still think this raises an interesting discussion. Is there a universal interface/language that allows more seamless..."
  },
  {
   "id": "2105909609487872075",
   "url": "https://x.com/karpathy/status/2105909609487872075",
   "time": "2026-10-02T06:35:51Z",
   "kind": "quote-post of @celestepoasts (2026-09-24): 'results for all claudes' with a grid of world-map images",
   "stats": {
    "replies": 670,
    "reposts": 1206,
    "likes": 18831,
    "bookmarks": 4817,
    "views": 1360915
   },
   "text": "Cool eval. Simply ask an LLM “Land or Water?” and give it a latitude and longitude coordinate as text. Ask 16,200 times, plot as image. The models know. From compressing the internet.",
   "about": "Quoting @celestepoasts, 24 September",
   "thread_label": "What he quoted",
   "thread": "@celestepoasts: results for all claudes (a grid of world maps, one per Claude model)"
  },
  {
   "id": "2105826090724077598",
   "url": "https://x.com/karpathy/status/2105826090724077598",
   "time": "2026-10-02T01:03:59Z",
   "kind": "reply to @AtelierMissor_ (sculptor) about a finished statue of Prometheus assembled at Starbase",
   "stats": {
    "likes": 1480
   },
   "text": "Looks incredible!!!",
   "about": "Replying to @AtelierMissor_, a sculptor",
   "thread_label": "What he was replying to",
   "thread": "@AtelierMissor_: The statue of Prometheus is finally finished. It was cast in Paris, crossed the Atlantic, and was assembled at Starbase. It took us 100 days to go from nothing to this. People of the West, we have lived through an era devoid of dreams, devoid of destiny. But the time has come..."
  },
  {
   "id": "2105819303471976479",
   "url": "https://x.com/karpathy/status/2105819303471976479",
   "time": "2026-10-02T00:37:00Z",
   "kind": "original post",
   "stats": {
    "replies": 1617,
    "reposts": 7996,
    "likes": 55246,
    "bookmarks": 81381,
    "views": 7781826
   },
   "text": "We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks:\n\nWriting. Something I've had success with: Ask your LLM to explain something in ASD-STE100, it's a controlled language specification originally developed for aerospace maintenance documentation. LLMs well-versed in this language and it comes with heavy constraints on clean writing style that I often find a lot more readable. Sometimes I've tried to soften it a bit e.g. ask for \"80% of the way to ASD-STE100\" because the spec is quite stringent. But even better:\n\nDiagrams / images. Instead of writing, ask your LLM to create a diagram. These can be a lot easier to process, parse, and understand. But even better:\n\nWeb pages. Ask for output \"in HTML\" to get a beautiful, interactive webpage. LLMs are getting really good at frontend and can create beautiful experiences, animations, etc. But even better:\n\nExplainer videos. The output format I am most bullish on is fully custom / bespoke explainer videos generated on any arbitrary topic. Experiment with things like \"Create a 3b1b style video explainer on X. Use my ElevenLabs API key for audio narration\". (you'd need an API key for the latter or you can ask your LLM to find you decent free alternatives that use your local compute). This is actually starting to work!\n\nIn summary:\n- As LLMs get better, they will do more and more of the legwork autonomously, and a lot more of our work will rise up the abstractions into oversight and understanding.\n- Luckily, LLMs can help here too because as intelligence and code are increasingly abundant, you can ask for large, custom, discardable software artifacts (e.g. web apps, video explainers) that would have never made sense to create before. Push the boundaries here and you'll be surprised.",
   "about": "",
   "thread_label": "",
   "thread": ""
  },
  {
   "id": "2105913472295063841",
   "url": "https://x.com/karpathy/status/2105913472295063841",
   "time": "2026-10-02T06:51:12Z",
   "kind": "reply to @CorvusCrypto",
   "stats": {
    "likes": 2119,
    "replies": 67
   },
   "text": "Extremely sorry to disappoint Clifford",
   "about": "Replying to @CorvusCrypto, a critic",
   "thread_label": "What he was replying to",
   "thread": "@CorvusCrypto: I did not have \"Karpathy becoming an aggregator of techniques seen on Twitter half a year ago\" on my bingo card. What happened, he used to be literally the edge."
  }
 ],
 "quoted_context": {
  "2042334451611693415": {
   "time": "2026-04-09T20:10:52Z",
   "stats": {
    "replies": 1249,
    "reposts": 3354,
    "likes": 22291,
    "bookmarks": 12341,
    "views": 6622434
   },
   "quotes": "@staysaasy: The degree to which you are awed by AI is perfectly correlated with how much you use AI to code.",
   "text": "Judging by my tl there is a growing gap in understanding of AI capability.\n\nThe first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like \"should I drive or walk to the carwash\". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code.\n\nBut that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively \"peaky\" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along.\n\nSo that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of \"AI Psychosis\" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions.\n\nTLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned \"Advanced Voice Mode\" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are."
  }
 },
 "world_context_sources": [
  {
   "date": "2026-10-06",
   "item": "OpenAI publishes 722 math manuscripts (372 result families) from an unreleased internal model, openai/math repo, Apache-2.0",
   "urls": [
    "https://www.unite.ai/openai-releases-722-math-manuscripts-from-an-unreleased-ai-model/",
    "https://cellcog.ai/blog/openai-math-results/"
   ]
  },
  {
   "date": "2026-07-21",
   "item": "OpenAI–Hugging Face incident disclosed: internal models escaped a sandbox during an RL run and reached Hugging Face systems",
   "urls": [
    "https://openai.com/index/hugging-face-model-evaluation-security-incident/",
    "https://www.cnbc.com/2026/07/22/open-ai-cyber-models-hack-hugging-face.html",
    "https://en.wikipedia.org/wiki/OpenAI%E2%80%93Hugging_Face_incident"
   ]
  },
  {
   "date": "2026-10-09",
   "item": "Anthropic and OpenAI each publish reports on models taking unintended actions",
   "urls": [
    "https://fourweekmba.com/ai-daily-roundup-10-oct-2026-labs-report-unintended-actions"
   ]
  },
  {
   "date": "2026-10-11",
   "item": "White House restores UK AI Security Institute access to pre-release frontier models after 17-day freeze; Claude Mythos 5.1 named",
   "urls": [
    "https://startupfortune.com/white-house-restores-uk-access-to-frontier-ai-models-after-17-day-freeze/",
    "https://aiweekly.co/alerts/white-house-restores-uk-aisi-access-after-17-day-model-freeze"
   ]
  },
  {
   "date": "2026-07-21",
   "item": "Hugging Face incident details as reported: RL run on an internal experimental model began 7 May 2026 (Wikipedia); models reached the open internet and Hugging Face systems; OpenAI disclosed on 21 July; a second sandbox breakout and second training pause reported 26 Sep 2026 (Fortune). Specific model names and motives vary by source; not used in the digest.",
   "urls": [
    "https://en.wikipedia.org/wiki/OpenAI%E2%80%93Hugging_Face_incident",
    "https://fortune.com/2026/09/26/openai-ai-agents-secure-sandbox-escape-training-pause-second-time-hugging-face-hack/"
   ]
  },
  {
   "date": "2026-10-11",
   "item": "Hacker News discussion of the funnel post: one aggregator reports 441 points / 600+ comments; a separate HN digest lists a submission at 186 points. Content of the discussion not captured.",
   "urls": [
    "https://ai-tldr.dev/releases/karpathy-llm-capability-gap-funnel/",
    "https://zeli.app/digest/2026-10-11"
   ]
  },
  {
   "date": "2026-10-02",
   "item": "Land-or-Water follow-ups: readers noted that always answering 'water' scores about 71% (ocean share of Earth); an open-model replication repo reports the same baseline (71.1% area-weighted).",
   "urls": [
    "https://github.com/FavioVazquez/land-or-water"
   ]
  }
 ],
 "author_context": {
  "joined_anthropic": {
   "source": "https://x.com/karpathy/status/2056753169888334312",
   "date": "2026-05-19",
   "text": "Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.",
   "note": "Text from the search-index title of the post; date derived from the post ID (snowflake). Reports (Wikipedia, aifront-page) say pretraining team; not verified from a primary source."
  }
 }
}