Jay Alammar
Writes Language Models & Co., a newsletter on large language models; known for The Illustrated Transformer, referenced in AI/ML courses at MIT, Stanford and Cornell.
Jay Alammar did not write this page. ppll compiled it from their own words, each linked to the source. Is this you? · Report a mistake
Latest Bluesky
What they believe
-
Running large batches of prompts through an LLM API is a capability unavailable to someone who only uses models through a chat playground interface.
Running thousands of prompts is one of the superpowers you do not get to use if all of your LLM usage is through a playground.
newsletter.languagemodels.coSaid 3 Nov 2025This on korrents
-
More reasoning gives a language model more time and compute to solve harder problems, but increases latency and compute cost.
On the one hand, more reasoning allows the model more time and compute to reason about a problem which helps it tackle more difficult problems. On the other hand, that comes at a cost of latency and compute.
newsletter.languagemodels.coSaid 19 Aug 2025This 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
16 November
12 November
14 October
7 October
6 October
5 October
1 October
29 September
25 September
24 September
23 September
22 September
14 September
3 November 2025
From one piece Inside NeurIPS 2025: The Year’s AI Research, Mapped 2 beliefs · newsletter.languagemodels.co
-
korrents.com
Delegating work to many AI agents at once is not yet a wise choice because of the costs involved.Their words
It's probably not even wise yet to delegate to many agents given the possible costs involved.
-
Their words
Running thousands of prompts is one of the superpowers you do not get to use if all of your LLM usage is through a playground.
13 October 2025
19 August 2025
From one piece The Illustrated GPT-OSS 2 beliefs · newsletter.languagemodels.co
-
Their words
Note that little of these architectural details is particularly novel. They're generally similar in the latest SoTA open source MoE models.
-
Their words
On the one hand, more reasoning allows the model more time and compute to reason about a problem which helps it tackle more difficult problems. On the other hand, that comes at a cost of latency and compute.
23 May 2025
10 February 2025
27 January 2025
From one piece The Illustrated DeepSeek-R1 2 beliefs · newsletter.languagemodels.co
-
Their words
Reasoning problems, in contrast to general chat or writing requests, can be automatically verified or labeled.
-
Their words
Modern base models have crossed a certain threshold of quality and capability (this base model was trained on 14.8 trillion high-quality tokens).
15 January 2025
14 January 2025
13 January 2025
12 December 2024
11 December 2024
10 December 2024
29 November 2024
28 November 2024
26 November 2024
16 October 2024
-
Their words
A central concept the general public should know is that language models are not merely text generators, but that they can form other systems (embedding, classification) that are useful for problem solving.
15 August 2024
13 November 2023
22 June 2023
21 May 2023
26 March 2023
Nothing matches.
Beliefs
Korrents What they believe 7 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
Running large batches of prompts through an LLM API is a capability unavailable to someone who only uses models through a chat playground interface.
Running thousands of prompts is one of the superpowers you do not get to use if all of your LLM usage is through a playground.
Inside NeurIPS 2025: The Year’s AI Research, Mapped Said 3 Nov 2025
Delegating work to many AI agents at once is not yet a wise choice because of the costs involved.
It's probably not even wise yet to delegate to many agents given the possible costs involved.
Inside NeurIPS 2025: The Year’s AI Research, Mapped Said 3 Nov 2025
More reasoning gives a language model more time and compute to solve harder problems, but increases latency and compute cost.
On the one hand, more reasoning allows the model more time and compute to reason about a problem which helps it tackle more difficult problems. On the other hand, that comes at a cost of latency and compute.
The Illustrated GPT-OSS Said 19 Aug 2025
Show 4 more
Open source AI models today largely share similar architectural approaches, with little genuinely novel architecture left to introduce.
Note that little of these architectural details is particularly novel. They're generally similar in the latest SoTA open source MoE models.
The Illustrated GPT-OSS Said 19 Aug 2025
Base large language models have now crossed a threshold of quality and capability significant enough to change what training methods are viable.
Modern base models have crossed a certain threshold of quality and capability (this base model was trained on 14.8 trillion high-quality tokens).
The Illustrated DeepSeek-R1 Said 27 Jan 2025
Reasoning problems, unlike general chat or writing tasks, can be automatically verified or labeled without human annotation.
Reasoning problems, in contrast to general chat or writing requests, can be automatically verified or labeled.
The Illustrated DeepSeek-R1 Said 27 Jan 2025
Language models are not merely text generators; they can also form other systems, such as embeddings and classifiers, that are useful for problem solving.
A central concept the general public should know is that language models are not merely text generators, but that they can form other systems (embedding, classification) that are useful for problem solving.
Our book, Hands-On Large Language Models, Is Now Out! Said 16 Oct 2024
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 2 things — 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
Uses Dream Studio
The final image was generated by Stable Diffusion using Dream Studio.
jalammar.github.io 1 Jan 2023
Liked Midjourney
The results of Midjourney have always stood out as especially beautiful.
jalammar.github.io 1 Jan 2023
Where they publish
Site jalammar.github.io Archive of his illustrated ML explainers; posting moved to the newsletter below in 2025.
Link verified 25 Sept 2026.
Newsletter Language Models & Co. Where new writing lands since 2025.
Open newsletter.languagemodels.co
Recent
- Our new book on AI agents is out, and it hit #1 16 Nov 2026 From the LLM to the agent, in 300+ hand-crafted figures. Our new O’Reilly book with Maarten Grootendorst, plus a giveaway of five ebook copies.
- Inside NeurIPS 2025: The Year’s AI Research, Mapped 3 Nov 2025 Using Cohere's Command A Reasoning and Embed 4 to Visualize the ~6,000 Papers Accepted to NeurIPS 2025
- The Illustrated GPT-OSS 19 Aug 2025 OpenAI releases their first open source LLM in six years
Show 9 more
- How Transformer LLMs Work [Free Course] 10 Feb 2025 Maarten and I teamed up with Andrew Ng to bring you the most accessible and up-to-date intro to Transformers yet!
- The Illustrated DeepSeek-R1 27 Jan 2025 A recipe for reasoning LLMs
- SWE-Bench authors reflect on the state of LLM agents at Neurips 2024 14 Jan 2025 The SWE-bench task measures AI agents on software engineering tasks at the level of a github issue.
- Our book, Hands-On Large Language Models, Is Now Out! 16 Oct 2024 We would love to know what you think!
- LLM Tokenizers, Semantic Search Course, And book update #2 13 Nov 2023 Let me tell you what large language model Tokenizers are, why they're fascinating, and how they're are under-explored.
- We're Writing a Book! "Hands-On Large Language Models" 22 Jun 2023 With lots of visuals explaining language understanding and generation with LLMs
- LLM University, Generative AI, AI Product Moats 21 May 2023 Enjoy these many awesome videos, articles, and and entire free course
- What a Time for Language Models 26 Mar 2023 New capabilities, new products, new possible futures
- Coming soon 12 Mar 2023 This is Language Models & Co..
Link verified 25 Sept 2026. Recent items update automatically from the channel.
x.com @JayAlammar Personal account.
Recent
- Our first book was about the LLM. Our second is about the agent. The gap between the two is the gap between "here's the code" and "I fixed the bug and the tests pass." I shared why we wrote An Illustrated Guide to AI Agents, what’s inside, and where to get it: https://t.co/kvxLxyOalH (Also giving away 5 ebook copies to commenters there.) 16 Nov 2026
- Woke up to this. An Illustrated Guide to AI Agents is #1 in Amazon's AI best sellers! @MaartenGr and I wrote it for a simple reason. Agents have already changed how software gets built, and soon how a lot of knowledge work gets done. The group of people who understand them should be far larger than it is today. We strongly believe that starts with building intuition. And intuition is best delivered with crisp visuals and a sequence of ideas curated so each one prepares the next. Then comes the code. Once the picture is clear, the code reads like a description of something you already understa… 12 Nov 2026
- How ML research shifted between 2025 and 2026: Beautiful Neurips research map from @askalphaxiv using @cohere Embed 5 embeddings! 14 Oct 2026
Show 10 more
- Figure 1-20 is the most important figure out of the 330+ in An Illustrated Guide to AI Agents. 1- It shows what today's LLM-based AI agents are composed of, including their user and environment loops. 2- It is also therefore a table of contents for Part 1: Agent Fundamentals, showing where each of the key components is explained in details. Read it here: Amazon: https://t.co/0dX0Js2IP8 O'Reilly: https://t.co/rrf0n2l3QG PDF e-book: 7 Oct 2026 ebooks.com
- Such a beautiful way to express math to humans. 6 Oct 2026
- Fascinating! Elephants talk with deep rumbles, alone or overlapping like a choir, and this research found the sounds carry a rough clue about what's happening (meeting up, staying in touch, deciding to move), whether it's one elephant or many. In a chorus, each elephant uses the kind of rumble that fits the situation, and in some situations, especially when coordinating plans, their voices drift to sound more alike as the conversation goes on. 5 Oct 2026
- "Music is a hidden exercise in arithmetic by a soul that does not know it is counting." ~Leibniz Schopenhauer: That is quite accurate from a lower standpoint and may be parodied in the following way to suit our higher view of music: Music is "a hidden exercise in metaphysics by a soul that does not know it is philosophising." 1 Oct 2026
- Friends in India! The @shroffpub edition is out, at a much friendlier price, in full color, and with free shipping in India. This is your chance to dig into the 330+ figures @MaartenGr and I crafted to explain the most important concepts in AI agents. 29 Sept 2026
- We stand on the shoulders of giants, @migueldeicaza. GNOME was my first Linux desktop back in college, and a lot of how I learned to program was on top of it. Hope you enjoy the book! 25 Sept 2026
- I have used Primus and it was a revelation of what the future of research can look like. Really excited to dig into this version scaling it up. 24 Sept 2026
- It exists! The first print copies of An Illustrated Guide to AI Agents have started arriving. The ebook is out now. Print ships from October 6, and pre-orders are open. If you read it, @MaartenGr and I would love to see a photo of your copy in the wild. And if you enjoy it, please leave a review and send us feedback! Amazon: 23 Sept 2026 amazon.com
- Join the us for a free masterclass on AI Agents next week! Together with @MaartenGr @joshuastarmer @ChrisJMcCormick @SerranoAcademy we have been cooking a gentle introduction to AI agents. 📅 Sept 29, 2026 22 Sept 2026
- Thank you for showing up, Amsterdam! It was delightful to be back at @pydataamsterdam and to share the stage with @MaartenGr, my co-author on An Illustrated Guide to AI Agents, a mere week after releasing the ebook version. Maarten covered the fundamentals of building agents and I talked about evaluating them. For most technical users, one key skill is the ability to read benchmark scores critically. I went through ten questions that often explain as much of the gap between two reported scores than the models themselves do. Some of these were hard-won lessons from building @cohere North Mini… 14 Sept 2026
Link verified 25 Sept 2026. Recent items update automatically from the channel.
Bluesky @jayalammar.bsky.social Personal account.
Recent
- Our first book was about the LLM. Our second is about the agent. The gap between the two is the gap between "here's the code" and "I fixed the bug." I shared why we wrote An Illustrated Guide to AI Agents, what’s inside, and where to get it: 16 Nov 2026 newsletter.languagemodels.co
- Inside NeurIPS 2025: The Year’s AI Research, Mapped New blog post! NeurIPS 2025 papers are out—and it’s a lot to take in. This visualization lets you explore the entire research landscape interactively, with clusters and @cohere.com LLM-generated explanations that make it easier to grasp. 3 Nov 2025
- The Illustrated Guide to AI Agents New book announcement! Thrilled that together with @maartengr.bsky.social , we're writing a new book titled “An Illustrated Guide to AI Agents” and published by @oreilly.bsky.social. 13 Oct 2025
Show 13 more
- The Illustrated GPT-OSS New post! A visual tour of the architecture, message formatting, and reasoning of the latest GPT. 19 Aug 2025 newsletter.languagemodels.co
- The legendary John Carmack at #upperbound: - Current AI focus is RL (with Richard Sutton) solving Atari games - Thinking in line with the Alberta Plan. - It was a misstep to start working too low-level (e.g., at the cuda level). I kept stepping up the stack chain until now in pytorch 23 May 2025
- The Illustrated DeepSeek-R1 Spent the weekend reading the paper and sorting through the intuitions. Here's a visual guide and the main intuitions to understand the model and the process that created it. 27 Jan 2025 newsletter.languagemodels.co
- Alphaxiv is an awesome way to discuss ML papers -- often with the authors themselves. Here's an intro and demo by @rajpalleti.bsky.social we shot at #Neurips2024 15 Jan 2025 youtube.com
- Floored that the repo for Hands-On Large Language Models is now at 3.6k Github stars! And excited that professors are starting to use the book to teach LLM courses. Reach out to us if we can be of assistance! And if you've liked the book, leave us a review on Amazon or Goodreads! 14 Jan 2025
- SWE-Bench has been one of the most important tasks measuring the progress of agents tackling software engineering in 2024. I caught up with two of its creators, @ofirpress.bsky.social and Carlos E. Jimenez to share their ideas on the state of LLM-backed agents. 13 Jan 2025 youtube.com
- Good morning #NeurIPS2024! Stop by the @cohere.com booth at 3PM today (Thursday) for a signed copy of Hands-On Large Language Models - it will introduce you to LLMs, their applications, as well as Cohere's Embed, Rerank, and Command-R models. Come early as quantities are limited! 12 Dec 2024
- I'll be in the Cohere #NeurIPS2024 booth most of this afternoon. Come say hi, ask questions, and yes, we're hiring! Tomorrow I'll be signing copies of my book at 3PM! Limited copies available! 11 Dec 2024
- Hi NeurIPS! Explore ~4,500 NeurIPS papers in this interactive visualization: jalammar.github.io/assets/neuri... (Click on a point to see the paper on the website) Uses @cohere.com models and @lelandmcinnes.bsky.social's datamapplot/umap to help make sense of the overwhelming scale of NeurIPS. 10 Dec 2024
- Sure to be thought provoking. The previous interview had fascinating thoughts on scifi (Dune Vs. Foundation), on AI competition for AI safety, and on successful scifi as a self-preventing prophecy. 29 Nov 2024
- Excited to see you all at NeurIPS this year! Let's hang! 28 Nov 2024
- Great insights 26 Nov 2024
- Join us for a panel on scientific communication Dec 4! 26 Nov 2024
Link verified 25 Sept 2026. Recent items update automatically from the channel.
GitHub jalammar Source for the site itself and notebooks (e.g. BERT visual guide).
Recent
- ecco — Explain, analyze, and visualize NLP language models. Ecco creates interactive visualizations directly in Jupyter notebooks explaining the behavior of Transformer-based language models (like GP… 15 Aug 2024
- ml 16 Feb 2023
- eccojs — Visualization components for ecco.ml 19 Dec 2020
Show 3 more
- simpleTensorFlowClassificationExample — SImple notebook and dataset to demonstrate classification in TensorFlow 28 Aug 2020
- NanodegreeProject 26 Nov 2019
- embeddings 2 Mar 2019
Link verified 25 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.








