-
Their words
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*.
Andrej Karpathy
Founding member of OpenAI and former director of AI at Tesla; creator of nanoGPT and the term "vibe coding".
Andrej Karpathy did not write this page. ppll compiled it from their own words, each linked to the source. Is this you? · Report a mistake
Latest x.com
What they believe
-
The gap in shared understanding of LLM capability is widening.
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*.
x.comSaid 11 Oct 2026This on korrents
-
LLMs will do more legwork autonomously
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.
x.comSaid 2 Oct 2026This on korrents
The bold line is korrents’ wording; the quote is theirs. These 2: what they wrote before what they said aloud, one per source, newest first.
Feed
As its own page →Hiding
11 October
2 October
From one piece @karpathy on X 2 beliefs · x.com
-
Their words
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.
-
korrents.com
LLMs will do more legwork autonomouslyTheir words
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.
12 September
7 September
3 August
2 August
21 July
23 June
12 June
9 June
-
Lovedrcmnd.app
Claude Fable 5Their words
This is a super exciting release - Claude Fable 5 is the same underlying model as Mythos but with added safeguards. The benchmarks are great and it's SOTA on everything by a margin but I'll add that *qualitatively* also, this is a major-version-bump-deserving step change forward
19 May
11 May
30 April
- x.com
- x.com
- BlogSequoia Ascent 2026 summary
10 April
9 April
-
Their words
TLDR the people in these two groups are speaking past each other.
-
Likedrcmnd.app
OpenAI CodexTheir words
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.
-
Likedrcmnd.app
Claude CodeTheir words
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.
4 April
-
Their words
Something I've been thinking about - I am bullish on people (empowered by AI) increasing the visibility, legibility and accountability of their governments.
28 March
-
Their words
The LLMs may elicit an opinion when asked but are extremely competent in arguing almost any direction. This is actually super useful as a tool for forming your own opinions, just make sure to ask different directions and be careful with the sycophancy.
26 March
25 March
-
Their words
One common issue with personalization in all LLMs is how distracting memory seems to be for the models. A single question from 2 months ago about some topic can keep coming up as some kind of a deep interest of mine with undue mentions in perpetuity.
20 March
From one piece Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI 8 beliefs · youtube.com
-
Their words
But okay, they shouldn't actually be enacting these ideas. There is a queue of ideas and there's maybe an automated scientist that comes up with ideas based on all the archive papers and GitHub repos and it funnels ideas in or researchers can contribute ideas, but it's a single queue and there is workers that pull items and they try them out.
-
Their words
And I've gotten to a certain point and I thought it was like fairly well tuned and then I let auto research go for like overnight and it came back with like tunings that I didn't see.
-
Their words
So the question is how do I refactor all the abstractions so that I'm not I have to arrange it once and hit go. The name of the game is how can you get more agents running for longer periods of time without your involvement doing stuff on your behalf?
+ 5 more
-
Their words
So I think the industry just has to reconfigure in so many ways that's like the customer is not the human anymore. It's like agents who are acting on behalf of humans and this refactoring will be will probably be substantial in a certain sense.
-
Their words
these apps that are on the app store for using these smart home devices, etc. Uh, these shouldn't even exist kind of in a certain sense. Like shouldn't it just be APIs and shouldn't agents be just using it directly?
-
Their words
But now it's not about flops, it's about tokens. So what is your token throughput and what token throughput do you command?
-
korrents.com
What limits you with coding agents is your own skill at stringing them together, not the capability of the models.Their words
I think to a large extent you feel like it's a skill issue. It's not that the capability is not there. It's that you just haven't found a way to string it together of what's available.
-
korrents.com
The way software gets built flipped in December: writing code yourself is now the exception, not the default.Their words
I kind of went from 80/20 of like, you know, uh to like 20/80 of writing code by myself versus just delegating to agents. And I don't even think it's 20/80 by now. I think it's a lot more than that. I don't think I've typed like a line of code probably since December basically.
16 March
12 February
3 January
19 December 2025
-
Lovedrcmnd.app
Nano BananaTheir words
Nano banana is one of the most incredible, paradigm-shifting models of 2025.
18 December 2025
-
Likedrcmnd.app
Dyson Big+QuietTheir words
Install HVAC filters, and/or get a standalone air purifier, e.g. I got the Dyson Big+Quiet because it's quite good, HEPA grade and doubles as a cool looking alien artifact in your room, but for the top top performance I'd get IQAir GC MultiGas XE
-
Recommendsrcmnd.app
IQAir GC MultiGas XETheir words
Install HVAC filters, and/or get a standalone air purifier, e.g. I got the Dyson Big+Quiet because it's quite good, HEPA grade and doubles as a cool looking alien artifact in your room, but for the top top performance I'd get IQAir GC MultiGas XE
-
Likedrcmnd.app
BobbyApprovedTheir words
Use apps like BobbyApproved and Ivy to scan the bar codes and get a lot of information about all ingredients and a score (I like and use both).
+ 6 more
-
Likedrcmnd.app
IvyTheir words
Use apps like BobbyApproved and Ivy to scan the bar codes and get a lot of information about all ingredients and a score (I like and use both).
-
Likedrcmnd.app
LocaleTheir words
Consider getting a home delivery service, e.g. I currently like and use Locale
-
Usesrcmnd.app
BluelandTheir words
Look for very few and simple ingredients and ideally "fragrance free" and "dye free". I currently use Blueland for all of these.
-
Recommendsrcmnd.app
Poison Like No OtherTheir words
Recommended reading: "Poison like no other" (on plastics), "In Defense of Food"
-
Recommendsrcmnd.app
In Defense of FoodTheir words
Recommended reading: "Poison like no other" (on plastics), "In Defense of Food"
-
Recommendsrcmnd.app
Poison SquadTheir words
(on food vs. "edible food-like substances"), "Poison Squad"
10 December 2025
29 November 2025
22 November 2025
18 November 2025
17 November 2025
12 November 2025
1 October 2025
21 August 2025
26 June 2025
1 May 2025
8 April 2025
24 March 2025
-
Dislikedrcmnd.app
AutoSleepTheir words
AutoSleep is basically a random number generator. Maybe there is a better app on Apple Watch for sleep tracking, but I haven't found it. Do not use.
18 August 2024
15 August 2024
Nothing matches.
Beliefs
Korrents What they believe 69 beliefs — each backed by an exact quote.
Each is a — compiled by korrents.com, not by them: the one-line wordings are korrents', the quotes are theirs.
Recent
The gap in shared understanding of LLM capability is widening.
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*.
@karpathy on X Said 11 Oct 2026
LLMs will do more legwork autonomously
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.
@karpathy on X Said 2 Oct 2026
As intelligence and code become increasingly abundant, people can ask for large, custom, discardable software artifacts that would never have made sense to create before.
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.
@karpathy on X Said 2 Oct 2026
Show 57 more
People disagreeing about AI capability are speaking past each other, because they are using models of very different tiers on very different kinds of task.
TLDR the people in these two groups are speaking past each other.
@karpathy on X Said 9 Apr 2026
AI will let citizens make their governments legible and accountable, reversing the historical direction in which only states could read society.
Something I've been thinking about - I am bullish on people (empowered by AI) increasing the visibility, legibility and accountability of their governments.
@karpathy on X Said 4 Apr 2026
Because an LLM can argue almost any direction competently, the right way to use one for forming an opinion is to make it argue every side.
The LLMs may elicit an opinion when asked but are extremely competent in arguing almost any direction. This is actually super useful as a tool for forming your own opinions, just make sure to ask different directions and be careful with the sycophancy.
@karpathy on X Said 28 Mar 2026
LLM memory as currently built is a distraction to the model: one old question keeps resurfacing as if it were a lasting interest.
One common issue with personalization in all LLMs is how distracting memory seems to be for the models. A single question from 2 months ago about some topic can keep coming up as some kind of a deep interest of mine with undue mentions in perpetuity.
@karpathy on X Said 25 Mar 2026
The way software gets built flipped in December: writing code yourself is now the exception, not the default.
I kind of went from 80/20 of like, you know, uh to like 20/80 of writing code by myself versus just delegating to agents. And I don't even think it's 20/80 by now. I think it's a lot more than that. I don't think I've typed like a line of code probably since December basically.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
What limits you with coding agents is your own skill at stringing them together, not the capability of the models.
I think to a large extent you feel like it's a skill issue. It's not that the capability is not there. It's that you just haven't found a way to string it together of what's available.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
The engineer's resource constraint has moved from flops to tokens: what matters now is the token throughput you command.
But now it's not about flops, it's about tokens. So what is your token throughput and what token throughput do you command?
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
The apps built to drive smart devices should not exist at all: the devices should expose APIs and agents should call them directly.
these apps that are on the app store for using these smart home devices, etc. Uh, these shouldn't even exist kind of in a certain sense. Like shouldn't it just be APIs and shouldn't agents be just using it directly?
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
Software's customer is no longer the human but the agent acting on their behalf, and the whole industry has to be refactored around that.
So I think the industry just has to reconfigure in so many ways that's like the customer is not the human anymore. It's like agents who are acting on behalf of humans and this refactoring will be will probably be substantial in a certain sense.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
The goal is not a better prompt but your own absence: arrange it once, hit go, and keep agents running for long stretches without you.
So the question is how do I refactor all the abstractions so that I'm not I have to arrange it once and hit go. The name of the game is how can you get more agents running for longer periods of time without your involvement doing stuff on your behalf?
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
An overnight automated search found tunings that two decades of hand-tuning experience had missed on an already well-tuned repository.
And I've gotten to a certain point and I thought it was like fairly well tuned and then I let auto research go for like overnight and it came back with like tunings that I didn't see.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
AI researchers should contribute ideas but stop enacting them: a single queue of ideas and automated workers should run the experiments.
But okay, they shouldn't actually be enacting these ideas. There is a queue of ideas and there's maybe an automated scientist that comes up with ideas based on all the archive papers and GitHub repos and it funnels ideas in or researchers can contribute ideas, but it's a single queue and there is workers that pull items and they try them out.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
Today's agents are a brilliant systems programmer and a ten-year-old at the same time, a combination you would never meet in a person.
I simultaneously feel like I'm talking to an extremely brilliant PhD student who's been like a systems programmer for their entire life and a 10-year-old.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
Reinforcement learning leaves models sharply jagged: on the rails of a verifiable domain they are superintelligent, off them everything meanders.
And so you're kind of like you're either on rails and you're part of the super intelligence circuits or you're not on rails and you're outside of the verifiable domains and suddenly everything kind of just like meanders.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
Capability does not generalise for free: a model that will move mountains on an agentic task still tells the same bad joke it told five years ago.
So even though the models have improved tremendously and if you give them an agentic task, they will just go for hours and move mountains for you. And then you ask for like a joke and it has a stupid joke. It's crappy joke from five years ago and it's because it's outside of the it's outside of the RL.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
Model intelligence should speciate like the animal kingdom rather than converge on one oracle that knows everything.
I do think we should expect more speciation in the intelligences. Um like, you know, the animal kingdom is extremely diverse in the brains that exist and there's lots of different niches of of nature and some animals have overdeveloped visual cortex or other part kind of parts and I think we we should be able to see more speciation and um you don't need like this oracle that knows everything.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
A swarm of untrusted volunteers on the internet could improve models and run circles around the frontier labs, because the earth has far more compute than they do.
a swarm of agents on the internet could collaborate to improve LLMs and could potentially even like run circles around frontier labs. Like who knows, you know? Um yeah, like maybe that's even possible. Like frontier labs have a huge amount of trusted compute but the earth is much bigger and has huge amount of untrusted compute.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
Demand for software will be extremely large precisely because software just got much cheaper to produce.
I do have like cautiously optimistic view of this in software engineering where I do think um it does seem to me like the demand for software will be extremely large. Um and it's just become a lot cheaper.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
Nobody inside a frontier lab can speak freely about AI, because there are things the organisation will not let them say and things it wants them to say.
You're you're not a completely free agent and you can't actually like be part of that conversation in a fully autonomous um free way. Like if you're inside one of the frontier labs. Like there's some things that you can't say. Uh and conversely there are some things that the organization wants you to say.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
Anyone outside a frontier lab will find their judgment about AI drifting, because the systems are opaque and what is coming is not visible from outside.
And I think if you're outside of that frontier lab, your your judgment fundamentally will start to drift because you're not part of the you know, what's coming down the line. And so I feel like my judgment will inevitably start to drift as well.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
Open models trailing the frontier by a few months is a good arrangement, because the industry needs a common working space it all has access to.
But I want there to be a thing that's behind and that uh is kind of like a common working space for intelligences that the entire industry has access to.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
Purely digital progress runs out: to keep learning, the machines will have to run experiments and ask the universe questions.
you're going to run out of things that you're going to do purely in the digital space. At some point you have to go to the universe and you have to ask it questions. Um you have to run an experiment and see what the universe tells you to get back to learn something.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
Stop writing documentation for people: a library should ship markdown for agents, which then explain it to whoever asks.
It used to be that you have documentation for other people who are going to use your library, but like you shouldn't do that anymore. Like you should have instead of HTML documents for humans, you have markdown documents for agents.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
An agent can understand a maximally simplified explanation and still be unable to come up with it — that gap is what is left of the expert's job.
so for example, micro GPT, like I asked I tried to get an agent to write micro GPT. So, I told it like try to boil down the simplest things. Like try to boil down my um neural network training to the simplest thing and it can't do it.
Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI Said 20 Mar 2026
The age of the IDE is not over; agents make it bigger, because humans now program at a higher level where the unit of interest is an agent rather than a file.
Expectation: the age of the IDE is over Reality: we’re going to need a bigger IDE (imo). It just looks very different because humans now move upwards and program at a higher level - the basic unit of interest is not one file but one agent. It’s still programming.
@karpathy on X Said 11 Mar 2026
Agentic organisations will be forkable in a way that classical companies never were.
You can’t fork classical orgs (eg Microsoft) but you’ll be able to fork agentic orgs.
@karpathy on X Said 11 Mar 2026
Outages at the frontier AI labs will become "intelligence brownouts" — the planet losing IQ points when the models stutter.
Intelligence brownouts will be interesting - the planet losing IQ points when frontier AI stutters.
@karpathy on X Said 11 Mar 2026
A tiny language model inventing a plausible-sounding name is the same phenomenon as a large one confidently stating a false fact.
microgpt “hallucinating” a name like “karia” is the same phenomenon as ChatGPT confidently stating a false fact.
microgpt Said 12 Feb 2026
LLM intelligence should not be understood through an animal analogy; it is a different kind of entity altogether
2025 is where I (and I think the rest of the industry also) first started to internalize the "shape" of LLM intelligence in a more intuitive sense. We're not "evolving/growing animals", we are "summoning ghosts".
2025 LLM Year in Review Said 19 Dec 2025
How verifiable a task is now predicts how automatable it is, the way specifiability predicted it in the 1980s.
The more a task/job is verifiable, the more amenable it is to automation in the new programming paradigm.
Verifiability Said 17 Nov 2025
A verifiable task can be optimised by reinforcement learning until a neural network performs it extremely well.
If a task/job is verifiable, then it is optimizable directly or via reinforcement learning, and a neural net can be trained to work extremely well.
Verifiability Said 17 Nov 2025
The best historical analogy for AI is not electricity or the industrial revolution but a new computing paradigm, because both are fundamentally about automating digital information processing.
AI has been compared to various historical precedents: electricity, industrial revolution, etc., I think the strongest analogy is that of AI as a new computing paradigm (Software 2.0) because both are fundamentally about the automation of digital information processing.
@karpathy on X Said 16 Nov 2025
Agents will take a decade rather than a year, because today's models are cognitively lacking in too many independent ways at once.
So like they don't have enough intelligence. They're not multimodal enough. They can't do computer use and all this kind of stuff. And they don't do a lot of things. You know, they don't have continual learning. You can't just tell them something and they'll remember it. And they're just cognitively lacking. And it's just not working. And I just think that it will take about a decade to work through all of those issues.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Computer-use agents had to wait for language models: without pre-trained representations the reward is too sparse to ever learn from.
And it just so turns out that um this was extremely early, way too early. so early that we shouldn't have been working on that, you know, uh because um if you're just stumbling your way around and keyboard mashing and mouse clicking and trying to get rewards in these environments, um your reward is too sparse and you just won't learn and you're going to burn a forest uh computing and you're never actually going to get something off the ground.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
We are not building animals, we are summoning ghosts: entities made by imitating human text rather than by evolution.
we're not doing training by evolution. Uh we're doing training by basically imitation of humans and the data that they've put on the internet. And so you end up with these like sort of ethereal spirit entities because they're fully digital and they're kind of like mimicking humans.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Humans barely use reinforcement learning for intelligence — what RL they do use goes into motor tasks, not problem solving.
a lot of what looks like learning is actually a lot more maturation of the brain and I think that actually very little reinforcement learning for animals and I think a lot of the reinforcement learning is actually like more like motor tasks. It's not intelligence tasks. So I actually kind of think humans don't actually like really use RL roughly speaking is what I would say.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Pre-training is a crappy evolution: the practically buildable substitute for the process that gave animals their built-in hardware.
So that's why I kind of call pre-training this kind of like crappy evolution. It's like the practically possible version with our technology and what we have available to us to get to a starting point where we can actually do things like reinforcement learning and so on.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
The knowledge a model soaks up in pre-training is holding it back; what we actually want is the intelligence with the knowledge stripped out.
And actually, you don't actually need or want the knowledge. I actually think that's probably actually holding back the neural networks overall because it's actually like getting them to rely on the knowledge a little too much sometimes.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
What is in a model's weights is a hazy recollection of things it read a year ago; what is in its context window is working memory.
anything that's in the weights it's kind of like a hazy recollection of what you read a year ago anything that you give it as a context uh at test time is directly in the working memory
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Language models have no equivalent of sleep: nothing takes the day's context, thinks it over, and distils it back into the weights.
These models don't really have this distillation phase um of taking what happened, analyzing it, obsessively thinking through it, um basically doing some kind of a synthetic data generation process and distilling it back back into the weights
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
You only really know something once you have built it — blog posts and slides leave gaps that building the thing does not.
So, don't write blog posts, don't do slides, don't do any of that. Like, build the code, arrange it, get it to work. It's the only way to go. Otherwise, you're missing knowledge.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Coding models are worst at exactly the thing an AI research explosion would need: code that has never been written before.
I think um yeah, they're not very good at code that hasn't never been written before maybe is like one way to put it, which is like what we're trying to achieve when we're building these models.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
The industry is pretending today's coding agents are amazing when what they produce is slop, and some of that pretending is fundraising.
And I kind of feel like the industry it's it's um it's over it's it's making too big of a jump and it's trying to pretend like this is amazing and it's not. It's slop and I think they're not coming to terms with it and maybe they're trying to fund raise or something like that.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Reinforcement learning is terrible, and it only looks good because everything we had before it was much worse.
reinforcement learning is a lot worse than I think the average person thinks reinforcement learning is terrible. It just so happens that uh everything that we had before is much worse
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
No person learns the way RL does: a human reviews which parts of an attempt were good instead of rewarding every step of a lucky one.
A human would never do this. Number one, a human would never do hundreds of rollouts. Uh number two, when a person sort of finds a solution, they will have a pretty complicated process of review of like, okay, I think these parts that I did well, these parts I did not do that well, I should probably do this or that.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Process supervision by an LLM judge cannot be run for long, because a judge with billions of parameters is gameable and RL will find its cracks.
the reason that I think this is kind of tricky is quite subtle. And it's the fact that anytime you use an LLM to assign a reward, those LLMs are giant things with billions of parameters and they're gameable.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Every sample a model gives you is silently collapsed: each one looks fine, but together they occupy a tiny sliver of the possible space.
all of the samples you get from models are silently collapsed. They're silently, this is not obvious if you look at any individual example of it. They occupy a very tiny manifold of the possible space of um sort of thoughts about content.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Humans collapse over the course of a life exactly as models do; children say shocking things because they have not overfitted yet.
I also think humans collapse over time. Uh I think this is uh again these analogies are surprisingly good but humans collapse during the course of their lives. This is why children have completely u you know they haven't overfit yet and they will say stuff that will shock you
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Models are enormous only because the internet is bad: most of that capacity does memory work, and a better dataset would need far less of it.
Um so I almost feel like because the internet is so terrible, we actually have to sort of like build really big models to compress all that. Uh most of that compression is memory work instead of like cognitive work. But what we really want is the cognitive part actually delete the memory
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Quietly dropping physical work from the definition of AGI is a major concession, not a clarification.
Now, number one, the first concession that people make all the time is they just take out all the physical stuff because we're just talking about digital knowledge work. I feel like that's a pretty major concession compared to the original definition which was like any task a human can do.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Hinton's prediction that radiology would stop being a job was wrong: radiologists are alive, well and growing in number.
so a good example recently was um Jeff Hinton's prediction that radiologists would not be a job anymore and this turned out to be very wrong in a bunch of ways right so radiologists are alive and well and growing even though computer vision is really really good at recognizing all the different things that they have to recognize
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Jobs will not be replaced outright: AIs will take most of the volume while humans supervise teams of them and handle the rest.
But even there I'm not actually looking at full automation yet. I'm looking for an autonomy slider and I almost expect that we are not going to instantly replace people. We're going to be swapping in AIs that do 80% of the volume. They delegate 20% of the volume to humans and humans are supervising teams of five AIs doing the call center work that's more rote.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Coding is the first job AI took because coding has always been text, which is the one thing these models are built to process.
so I think there's there's an interesting point here because I do believe coding is like the perfect first thing for uh for a for uh these LLMs and uh agents and that's because coding has always fundamentally uh worked around text.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
We are already in an intelligence explosion and have been for decades, so AI will show up as more of the same exponential rather than a break in it.
I do, but it's business as usual because we're we're in an intelligence explosion already and have been for decades. And when you look at GDP, it's basically the GDP curve that is an exponential weighted sum over so many aspects of the industry. Everything is gradually being automated has been for hundreds of years.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
The most likely bad outcome is not a takeover but a gradual loss of control and understanding, as fewer and fewer people understand what we have layered everywhere.
and we'll gradually layer all this stuff everywhere and there will be fewer and fewer people who understand it and that there will be a sort of this like scenario of a gradual loss of control and understanding of what's happening that to me seems most likely outcome of how all this stuff will go down.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Getting from a demo to a product is a march of nines, and each additional nine costs the same amount of work as the one before it.
I think basically what takes the long amount of time and the way to think about it is that it's a march of nines and every single nine is a constant amount of work.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
Self-driving has not removed the human from the car; it has moved them somewhere the passenger cannot see them.
I think some of the times they are but they're certainly involved and there are people and in some sense we haven't actually removed the person we've like moved them to somewhere where we can't see them.
Andrej Karpathy — “We’re summoning ghosts, not building animals” Said 17 Oct 2025
What is a korrent?
A korrent is a belief a person has stated in their own words: one sentence stating the claim, backed by a quote and a source, kept at korrents.com.
Under a name here, the quoted block is what they actually said. The korrent beneath it is the claim those words support, in korrents' wording — tap it to see the record, its source, and who else holds it.
Nobody here wrote their own korrents. They are compiled from public statements, and a person can change their mind, which is recorded too.
Recommends rcmnd
rcmnd What they recommend 34 things, 35 mentions — each backed by an exact quote.
Compiled by rcmnd.app, not by them: the quotes are theirs, and each links to the post it was copied from.
Recent
Loved Claude Fable 5
This is a super exciting release - Claude Fable 5 is the same underlying model as Mythos but with added safeguards. The benchmarks are great and it's SOTA on everything by a margin but I'll add that *qualitatively* also, this is a major-version-bump-deserving step change forward
x.com 9 Jun 2026
Liked Claude Code
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.
x.com 9 Apr 2026
Liked OpenAI Codex
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.
x.com 9 Apr 2026
Show 32 more
Loved Nano Banana
Nano banana is one of the most incredible, paradigm-shifting models of 2025.
karpathy.bearblog.dev 19 Dec 2025
Liked Claude Code
Claude Code (CC) emerged as the first convincing demonstration of what an LLM Agent looks like - something that in a loopy way strings together tool use and reasoning for extended problem solving.
karpathy.bearblog.dev 19 Dec 2025
Recommends Poison Squad
(on food vs. "edible food-like substances"), "Poison Squad"
karpathy.bearblog.dev 18 Dec 2025
Recommends In Defense of Food
Recommended reading: "Poison like no other" (on plastics), "In Defense of Food"
karpathy.bearblog.dev 18 Dec 2025
Recommends Poison Like No Other
Recommended reading: "Poison like no other" (on plastics), "In Defense of Food"
karpathy.bearblog.dev 18 Dec 2025
Uses Blueland
Look for very few and simple ingredients and ideally "fragrance free" and "dye free". I currently use Blueland for all of these.
karpathy.bearblog.dev 18 Dec 2025
Liked Locale
Consider getting a home delivery service, e.g. I currently like and use Locale
karpathy.bearblog.dev 18 Dec 2025
Liked Ivy
Use apps like BobbyApproved and Ivy to scan the bar codes and get a lot of information about all ingredients and a score (I like and use both).
karpathy.bearblog.dev 18 Dec 2025
Liked BobbyApproved
Use apps like BobbyApproved and Ivy to scan the bar codes and get a lot of information about all ingredients and a score (I like and use both).
karpathy.bearblog.dev 18 Dec 2025
Recommends IQAir GC MultiGas XE
Install HVAC filters, and/or get a standalone air purifier, e.g. I got the Dyson Big+Quiet because it's quite good, HEPA grade and doubles as a cool looking alien artifact in your room, but for the top top performance I'd get IQAir GC MultiGas XE
karpathy.bearblog.dev 18 Dec 2025
Liked Dyson Big+Quiet
Install HVAC filters, and/or get a standalone air purifier, e.g. I got the Dyson Big+Quiet because it's quite good, HEPA grade and doubles as a cool looking alien artifact in your room, but for the top top performance I'd get IQAir GC MultiGas XE
karpathy.bearblog.dev 18 Dec 2025
Disliked AutoSleep
AutoSleep is basically a random number generator. Maybe there is a better app on Apple Watch for sleep tracking, but I haven't found it. Do not use.
karpathy.bearblog.dev 24 Mar 2025
Mixed on 8Sleep Pod 4 Ultra
I didn't find 8Sleep to be very reliable in its sleep tracking. The scores don't make as much sense to me when I wake up, and as we saw above they don't correlate very strongly with Whoop or Oura.
karpathy.bearblog.dev 24 Mar 2025
Recommends Oura
Summarizing all of that into my advice right now: Get Whoop for 9.5/10, reliable, convenient sleep tracking with an excellent app (once you get to know it a bit). Get Oura for 10/10 tracking, if you're ok with the ring form factor.
karpathy.bearblog.dev 24 Mar 2025
Recommends Whoop
Summarizing all of that into my advice right now: Get Whoop for 9.5/10, reliable, convenient sleep tracking with an excellent app (once you get to know it a bit). Get Oura for 10/10 tracking, if you're ok with the ring form factor.
karpathy.bearblog.dev 24 Mar 2025
Uses Apple Notes
I maintain one single text note in the Apple Notes app just called "notes".
karpathy.bearblog.dev 19 Mar 2025
Liked Little Snitch
I like and use The Little Snitch
karpathy.bearblog.dev 17 Mar 2025
Recommends pi-hole
For more advanced users who like to tinker, pi-hole is the physical alternative.
karpathy.bearblog.dev 17 Mar 2025
Liked NextDNS
You can block ads by blocking entire domains at the DNS level. I like and use NextDNS
karpathy.bearblog.dev 17 Mar 2025
Recommends Mullvad VPN
If you wish to hide your IP/location to services, you can do so via VPN indirection. I recommend Mullvad VPN
karpathy.bearblog.dev 17 Mar 2025
Uses Proton Mail
I still use gmail just due to sheer convenience, but I've started to partially use Proton Mail
karpathy.bearblog.dev 17 Mar 2025
Liked Virtual Post Mail
I currently use Earth Class Mail but tbh I'm a bit embarrassed by that and I'm looking to switch to Virtual Post Mail
karpathy.bearblog.dev 17 Mar 2025
Mixed on Earth Class Mail
I currently use Earth Class Mail but tbh I'm a bit embarrassed by that and I'm looking to switch to Virtual Post Mail
karpathy.bearblog.dev 17 Mar 2025
Liked Privacy.com
I like and use privacy.com to mint new credit cards for every single transaction or merchant.
karpathy.bearblog.dev 17 Mar 2025
Recommends Brave Search
Brave Search is a privacy-first search engine with its own index, unlike e.g. Duck Duck Go which is basically a nice skin for Bing
karpathy.bearblog.dev 17 Mar 2025
Recommends Brave Browser
Recommends Signal
I recommend Signal instead of text messages because it end-to-end encrypts all your communications.
karpathy.bearblog.dev 17 Mar 2025
Recommends YubiKey
basically I strongly recommend the use of 2-3 YubiKeys to dramatically strengthen your digital security.
karpathy.bearblog.dev 17 Mar 2025
Liked 1Password
Combine this with a browser extension to create and Autofill them super fast. For example, I use and like 1Password
karpathy.bearblog.dev 17 Mar 2025
Liked Advanced Nutrition and Human Metabolism
undergoing some of these experiments was incredibly fun and a great excuse to study a number of textbooks on biochemistry (I liked "Molecular Biology of the Cell"), biology (I liked Campbell's Biology), human nutrition (I liked "Advanced Nutrition and Human Metabolism"), etc.
karpathy.github.io 11 Jun 2020
What others say about Advanced Nutrition and Human Metabolism →
Liked Campbell's Biology
undergoing some of these experiments was incredibly fun and a great excuse to study a number of textbooks on biochemistry (I liked "Molecular Biology of the Cell"), biology (I liked Campbell's Biology), human nutrition (I liked "Advanced Nutrition and Human Metabolism"), etc.
karpathy.github.io 11 Jun 2020
Liked Molecular Biology of the Cell
undergoing some of these experiments was incredibly fun and a great excuse to study a number of textbooks on biochemistry (I liked "Molecular Biology of the Cell"), biology (I liked Campbell's Biology), human nutrition (I liked "Advanced Nutrition and Human Metabolism"), etc.
karpathy.github.io 11 Jun 2020
Where they publish
Site karpathy.ai Bio, talks and papers.
An academic-style homepage: bio, selected talks and papers, and links out to the projects and the rest of these channels.
Link verified 1 Sept 2026.
Blog Andrej Karpathy blog The long technical write-ups, including the ones people still cite.
The older long-form technical write-ups, including ones still cited years later — "The Unreasonable Effectiveness of Recurrent Neural Networks" among them.
Recent
- microgpt 12 Feb 2026 This is a brief guide to my new art project microgpt, a single file of 200 lines of pure Python with no dependencies that trains and inferences a GPT. This file contains the full algorithmic content of what is needed: d…
- Deep Neural Nets: 33 years ago and 33 years from now 14 Mar 2022 The Yann LeCun et al. (1989) paper Backpropagation Applied to Handwritten Zip Code Recognition is I believe of some historical significance because it is, to my knowledge, the earliest real-world application of a neural…
- A from-scratch tour of Bitcoin in Python 21 Jun 2021 I find blockchain fascinating because it extends open source software development to open source + state. This seems to be a genuine/exciting innovation in computing paradigms; We don’t just get to share code, we get to…
Show 7 more
- Short Story on AI: Forward Pass 27 Mar 2021 The inspiration for this short story came to me while reading Kevin Lacker’s Giving GPT-3 a Turing Test. It is probably worth it (though not required) to skim this post to get a bit of a background on some of this story…
- Biohacking Lite 11 Jun 2020 Throughout my life I never paid too much attention to health, exercise, diet or nutrition. I knew that you’re supposed to get some exercise and eat vegetables or something, but it stopped at that (“mom said”-) level of…
- A Recipe for Training Neural Networks 25 Apr 2019 Some few weeks ago I posted a tweet on “the most common neural net mistakes”, listing a few common gotchas related to training neural nets. The tweet got quite a bit more engagement than I anticipated (including a webin…
- (started posting on Medium instead) 20 Jan 2018 The current state of this blog (with the last post 2 years ago) makes it look like I’ve disappeared. I’ve certainly become less active on blogs since I’ve joined Tesla, but whenever I do get a chance to post something I…
- A Survival Guide to a PhD 7 Sept 2016 This guide is patterned after my “Doing well in your courses”, a post I wrote a long time ago on some of the tips/tricks I’ve developed during my undergrad. I’ve received nice comments about that guide, so in the same s…
- Deep Reinforcement Learning: Pong from Pixels 31 May 2016 This is a long overdue blog post on Reinforcement Learning (RL). RL is hot! You may have noticed that computers can now automatically learn to play ATARI games (from raw game pixels!), they are beating world champions a…
- Short Story on AI: A Cognitive Discontinuity. 14 Nov 2015 The idea of writing a collection of short stories has been on my mind for a while. This post is my first ever half-serious attempt at a story, and what better way to kick things off than with a story on AI and what that…
Link verified 1 Sept 2026. Recent items update automatically from the channel.
Blog karpathy.bearblog.dev Current blog, and the source the quotes on this site were compiled from.
The current home for new writing, and the source most of the quotes on this site were compiled from. Sparse and careful — a post here tends to be a considered position, not a reaction.
Recent
- Sequoia Ascent 2026 summary 30 Apr 2026 Summary of my talk at Sequoia Ascent
- Chemical hygiene 22 Dec 2025 An evolving guide of protecting your health from a pricemaxxing industry.
- 2025 LLM Year in Review 20 Dec 2025 2025 Year in Review of LLM paradigm changes
Show 7 more
- Auto-grading decade-old Hacker News discussions with hindsight 10 Dec 2025 A vibe coding thought exercise on what it might look like for LLMs to scour human historical data at scale and in retrospect.
- The space of minds 29 Nov 2025 On the space of minds and the optimizations that give rise to them.
- Verifiability 17 Nov 2025 The impact of verifiability on the jagged frontier of LLMs
- Animals vs Ghosts 1 Oct 2025 Today's frontier LLM research is not about building animals. It is about summoning ghosts. And a bit more on Sutton's Dwarkesh pod.
- Vibe coding MenuGen 1 May 2025 Work log of vibe coding menugen app
- Power to the people: How LLMs flip the script on technology diffusion 8 Apr 2025
- Finding the Best Sleep Tracker 25 Mar 2025
Link verified 1 Sept 2026. Recent items update automatically from the channel.
x.com @karpathy Where the threads that get quoted everywhere start.
Long threads on AI that get screenshotted everywhere — this is where "vibe coding" was coined. The fastest-moving of their channels, and the one where ideas appear before they harden into posts.
Recent
- 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. - Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all. - 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… 11 Oct 2026
- 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. 2 Oct 2026
- We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks: Writing. 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: Diagrams / images… 2 Oct 2026
Show 17 more
- I love this and really hope we can come together as an industry and make it happen. 12 Sept 2026
- We're starting to leave the territory where you'd test an LLM by e.g. "create an svg of pelican on a bicycle". As one idea to generalize it, I was interested what Opus 5 would do if I gave it the first paragraph of the Lord of the Rings, a 1M token budget (~$10) and asked for three js render of it. Opus went off for ~2 hours and wrote 5500 lines of code that (procedurally) rendered the story. It's kind of janky but fun. But it's a bit mindboggling that the LLM has to place and orchestrate various polygon assets in (x,y,z) coordinates and write code that animates it all, and that it even does… 2 Aug 2026
- One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. In these cases I like to lean back, switch to /voice and just ramble for like 10 minutes, total mess, anything goes, full stream of consciousness. Sometimes I declare it up top, something like "switching to speech recognition sorry for any typos...". Sometimes I turn it into a small interview of a few turns. But I find that the LLMs are somehow very good at reconstructing long incoherent rambles and ofte… 21 Jul 2026
- This is a new paradigm for interacting with Claude that is significantly more "inline" with all the other human activity org-wide. Once you do all of the under the hood engineering work to make this "just work" (e.g. across tools, integrations, compute environments, memory, security, etc.), Claude basically joins the team in a seamless way - you can talk to it as you would talk to a person and it can help with a very large variety of workloads. Imo this is the 3rd major redesign of LLM UIUX. The first paradigm was that the LLM is a website you go to, the second was that it is an app you downl… 23 Jun 2026
- In awe of SpaceX and its story - past, present and the future. You can think about it in 10+ different ways and continue re-blowing your mind in circles. Huge congrats to the team! 🚀 12 Jun 2026
- This is a super exciting release - Claude Fable 5 is the same underlying model as Mythos but with added safeguards. The benchmarks are great and it's SOTA on everything by a margin but I'll add that *qualitatively* also, this is a major-version-bump-deserving step change forward (imo of the same order as Claude 4.5 was in November), peaking especially for long problem-solving sessions on very difficult problems. You can give it a lot more ambitious tasks than what you're used to, the model "gets it" and it will just go, and it's never felt this tempting to stop looking at the code at all (but… 9 Jun 2026
- 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. 19 May 2026
- This works really well btw, at the end of your query ask your LLM to "structure your response as HTML", then view the generated file in your browser. I've also had some success asking the LLM to present its output as slideshows, etc. More generally, imo audio is the human-preferred input to AIs but vision (images/animations/video) is the preferred output from them. Around a ~third of our brains are a massively parallel processor dedicated to vision, it is the 10-lane superhighway of information into brain. As AI improves, I think we'll see a progression that takes advantage: 1) raw text (hard… 11 May 2026
- This is the the quote I've been citing a lot recently. 30 Apr 2026
- Fireside chat at Sequoia Ascent 2026 from a ~week ago. Some highlights: The first theme I tried to push on is that LLMs are about a lot more than just speeding up what existed before (e.g. coding). Three examples of new horizons: 1. menugen: an app that can be fully engulfed by LLMs, with no classical code needed: input an image, output an image and an LLM can natively do the thing. 2. install .md skills instead of install .sh scripts. Why create a complex Software 1.0 bash script for e.g. installing a piece of software if you can write the installation out in words and say "just show this to… 30 Apr 2026
- Judging by my tl there is a growing gap in understanding of AI capability. The 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 roun… 9 Apr 2026
- Farzapedia, personal wikipedia of Farza, good example following my Wiki LLM tweet. I really like this approach to personalization in a number of ways, compared to "status quo" of an AI that allegedly gets better the more you use it or something: 1. Explicit. The memory artifact is explicit and navigable (the wiki), you can see exactly what the AI does and does not know and you can inspect and manage this artifact, even if you don't do the direct text writing (the LLM does). The knowledge of you is not implicit and unknown, it's explicit and viewable. 2. Yours. Your data is yours, on your loca… 4 Apr 2026
- Something I've been thinking about - I am bullish on people (empowered by AI) increasing the visibility, legibility and accountability of their governments. Historically, it is the governments that act to make society legible (e.g. "Seeing like a state" is the common reference), but with AI, society can dramatically improve its ability to do this in reverse. Government accountability has not been constrained by access (the various branches of government publish an enormous amount of data), it has been constrained by intelligence - the ability to process a lot of raw data, combine it with doma… 4 Apr 2026
- Wow, this tweet went very viral! I wanted share a possibly slightly improved version of the tweet in an "idea file". The idea of the idea file is that in this era of LLM agents, there is less of a point/need of sharing the specific code/app, you just share the idea, then the other person's agent customizes & builds it for your specific needs. So here's the idea in a gist format: https://t.co/NlAfEJjtJV You can give this to your agent and it can build you your own LLM wiki and guide you on how to use it etc. It's intentionally kept a little bit abstract/vague because there are so many directio… 4 Apr 2026
- LLM Knowledge Bases Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So: Data ingest: I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki… 2 Apr 2026
- New supply chain attack this time for npm axios, the most popular HTTP client library with 300M weekly downloads. Scanning my system I found a use imported from googleworkspace/cli from a few days ago when I was experimenting with gmail/gcal cli. The installed version (luckily) resolved to an unaffected 1.13.5, but the project dependency is not pinned, meaning that if I did this earlier today the code would have resolved to latest and I'd be pwned. It's possible to personally defend against these to some extent with local settings e.g. release-age constraints, or containers or etc, but I thin… 31 Mar 2026 stepsecurity.io
- - Drafted a blog post - Used an LLM to meticulously improve the argument over 4 hours. - Wow, feeling great, it’s so convincing! - Fun idea let’s ask it to argue the opposite. - LLM demolishes the entire argument and convinces me that the opposite is in fact true. - lol The LLMs may elicit an opinion when asked but are extremely competent in arguing almost any direction. This is actually super useful as a tool for forming your own opinions, just make sure to ask different directions and be careful with the sycophancy. 28 Mar 2026
Link verified 1 Sept 2026. Recent items update automatically from the channel.
YouTube Andrej Karpathy The multi-hour build-it-from-scratch lectures.
The multi-hour build-it-from-scratch lectures — the "Neural Networks: Zero to Hero" course, building GPT line by line — aimed at people who want to follow every step, not just the summary.
Link verified 1 Sept 2026.
Podcast As a guest Audio interviews they chose to sit for — chosen, not exhaustive.
Their own voice by their own participation: podcast conversations they agreed to, on the hosts' own shows. One for now rather than a list padded with episodes nobody checked.
Links
-
Andrej Karpathy — AGI is still a decade away
Dwarkesh Patel
17 Oct 2025
Link verified 1 Sept 2026.
-
Andrej Karpathy — AGI is still a decade away
Dwarkesh Patel
GitHub karpathy nanoGPT, micrograd, and the teaching repos.
The teaching repos that pair with the lectures — nanoGPT, micrograd, llm.c and the rest. Small, readable codebases meant to be studied, not just run.
Recent
- nanochat — The best ChatGPT that $100 can buy. 7 Sept 2026
- micrograd — A tiny scalar-valued autograd engine and a neural net library on top of it with PyTorch-like API 3 Aug 2026
- karpathy.github.io — my blog 10 Apr 2026
Show 16 more
- autoresearch — AI agents running research on single-GPU nanochat training automatically 26 Mar 2026
- jobs — A research tool for visually exploring Bureau of Labor Statistics Occupational Outlook Handbook data. This is not a report, a paper, or a serious economic publication — it is a development too… 16 Mar 2026
- rustbpe — The missing tiktoken training code 3 Jan 2026
- hn-time-capsule — Analyzing Hacker News discussions from a decade ago in hindsight with LLMs 10 Dec 2025
- llm-council — LLM Council works together to answer your hardest questions 22 Nov 2025
- reader3 — Quick illustration of how one can easily read books together with LLMs. It's great and I highly recommend it. 18 Nov 2025
- nanoGPT — The simplest, fastest repository for training/finetuning medium-sized GPTs. 12 Nov 2025
- rendergit — Render any git repo into a single static HTML page for humans or LLMs 21 Aug 2025
- llm.c — LLM training in simple, raw C/CUDA 26 Jun 2025
- nn-zero-to-hero — Neural Networks: Zero to Hero 18 Aug 2024
- minGPT — A minimal PyTorch re-implementation of the OpenAI GPT (Generative Pretrained Transformer) training 15 Aug 2024
- build-nanogpt — Video+code lecture on building nanoGPT from scratch 13 Aug 2024
- llama2.c — Inference Llama 2 in one file of pure C 6 Aug 2024
- LLM101n — LLM101n: Let's build a Storyteller 1 Aug 2024
- minbpe — Minimal, clean code for the Byte Pair Encoding (BPE) algorithm commonly used in LLM tokenization. 1 Jul 2024
- makemore — An autoregressive character-level language model for making more things 4 Jun 2024
Link verified 1 Sept 2026. Recent items update automatically from the channel.
About the English under a post
Some people here publish in a language other than English. Where they do, this site shows a machine translation beneath the post, in this typeface — the site's own, not theirs.
The post itself is never changed, moved or hidden: what is set in the serif above is exactly what the person published, and it is what to quote them on. A translation can be wrong in ways that matter, especially about tone.
Only the post's own words are translated. A quoted post, a linked article and a belief on korrents.com are left in their original language.