| 1 | Jun 5, '25 | some thoughts on human-ai relationships and how we're approaching them at openai it's a long blog post -- tl;dr we build models to serve people first. as more people feel increasin | 219× | 1.6M | 3.5k | 661 | 3.3k |
| 2 | Mar 27, '25 | // i lead model behavior at openai, and wanted to share some thoughts & nuance that went into setting policy for 4o image generation. features capital letters (!) bc i published it | 162× | 1.2M | 2.5k | 349 | 1.1k |
| 3 | Sep 26, '24 | at least i can now tweet about how hot and brilliant she is without getting into hr trouble or sounding like a corporate shill | 157× | 1.2M | 3.5k | 65 | 373 |
| 4 | Apr 6 | i shared this note on slack: | 156× | 1.2M | 4.8k | 300 | 1.3k |
| 5 | Nov 22, '23 | thank god | 150× | 1.1M | 1.6k | 28 | 95 |
| 6 | Sep 5, '25 | 🧪 i’m starting oai labs: a research-driven group focused on inventing and prototyping new interfaces for how people collaborate with ai. i’m excited to explore patterns that move | 84.4× | 631k | 2.1k | 88 | 717 |
| 7 | Nov 19, '23 | 💕 | 65.3× | 488k | 1.2k | 26 | 35 |
| 8 | Feb 23, '25 | Concerning (especially because i really dig grok3 as a model) | 63.9× | 478k | 1.6k | 57 | 245 |
| 9 | May 13, '24 | twitter's going to love finding out who leads research for personality @ oai | 56.7× | 424k | 565 | 14 | 142 |
| 10 | Jan 18, '25 | i find mishearing 'agents' as 'asians' way more entertaining "a swarm of asians working 24/7 while you sleep" "the future is asians" "you'll be able to deploy millions of asians in | 54× | 403k | 3.9k | 163 | 427 |
| 11 | Aug 29, '25 | today feels surreal. on the same day i was included in the time100 ai list, we shared internally that i’m transitioning from leading model behavior to begin something new at openai | 49.8× | 372k | 1.9k | 62 | 308 |
| 12 | Sep 28, '24 | i guess hr had bigger fish to fry | 47.4× | 355k | 928 | 8 | 97 |
| 13 | Mar 1, '23 | 🎉 The ChatGPT API is here! The energy, execution speed, and insights from developers during the alpha period have been unparalleled and made work _so_ fun. Read more about today's | 46.4× | 347k | 592 | 68 | 129 |
| 14 | Nov 16, '23 | If you've ever wondered: (1) how research and product collaborate at OpenAI, or (2) how ChatGPT should respond when told to be a cat 🐈 – Check out this talk @barret_zoph and I gav | 44× | 329k | 513 | 27 | 407 |
| 15 | Mar 1, '23 | $2 can get you 1M tokens – or approximately 750,000 words – on the ChatGPT API! | 34.5× | 258k | 960 | 74 | 227 |
| 16 | Aug 16, '25 | i think everyone in ai should think about what their "line" is. where if your company knowingly crosses that line and won't walk it back, you'll walk away this line is personal, wi | 32.7× | 245k | 1.2k | 91 | 255 |
| 17 | Apr 1, '25 | 🚫 on imagegen refusals we hear you on the current state of refusals not being up to far (amidst the transition discussed below)! favor: can you please reply with screenshots of im | 31.8× | 237k | 530 | 28 | 142 |
| 18 | Apr 8, '22 | ✨ We launched DALL•E 2 yesterday! 🎨🖌🤖 Working on this has been dreamy as a PM, because it’s not often that you get to ship what feels like magic. Nor is it often that your teamm | 30.4× | 0 | 3.3k | 316 | 370 |
| 19 | Feb 13, '24 | 📝 we just launched a small experiment for memory on ChatGPT. how it works - it's quite similar to custom instructions, except chatgpt is the one driving it (like auto vs. stick sh | 25.4× | 190k | 681 | 82 | 274 |
| 20 | Oct 7, '25 | hi, come build new interfaces with me @OpenAI ! computer : gui :: ai : ? | 25.4× | 190k | 946 | 34 | 474 |
| 21 | Apr 30, '25 | hi | 24.1× | 180k | 863 | 21 | 97 |
| 22 | Nov 20, '23 | 700+ now | 23.3× | 174k | 521 | 33 | 23 |
| 23 | Dec 22, '23 | this is what keeps me at OpenAI: specifically, our C-suite in particular has zero tolerance for bullshit. this empowers me to call out BS whenever I see it, and prevented me from q | 22.4× | 168k | 632 | 27 | 181 |
| 24 | Nov 25, '23 | impossible to plan a wedding without knowing when devday 2024 is | 21.9× | 164k | 570 | 4 | 20 |
| 25 | Sep 30, '25 | a day in the life of an openai employee 💕 | 21.7× | 163k | 676 | 23 | 121 |
| 26 | May 13, '24 | btw these are all _actually_ live demos! | 20.6× | 154k | 632 | 22 | 21 |
| 27 | Sep 12, '24 | 🍓 there's a lot of o1 hype on my feed, so i'm worried that it might be setting the wrong expectations what o1 is: the first reasoning model that shines in really hard tasks, and i | 20.5× | 153k | 1.2k | 122 | 215 |
| 28 | Nov 23, '23 | likely timeline: > 7:04 pm Elon DMs an on-call iOS engineer to change it > eng adds designer and goes "ok but that'll lead to higher news clickthroughs which I thought you di | 19.6× | 146k | 463 | 27 | 64 |
| 29 | Jun 12, '25 | not 4o, but 4ever 💍 | 19.2× | 144k | 2.5k | 18 | 136 |
| 30 | Feb 13, '25 | 🔪 chatgpt refusals & warnings 🙅♂️🙅♀️ has chatgpt ever refused to give you what you want for no good reason? or reasons you disagree with? please let us know and we'll do our b | 18.4× | 137k | 381 | 32 | 85 |
| 31 | Nov 21, '23 | you just gotta ship through it | 17.8× | 133k | 387 | 21 | 11 |
| 32 | Nov 12, '24 | View post on X | 17.4× | 130k | 526 | 2 | 92 |
| 33 | Mar 2, '23 | 🐱 Attention everyone! 🐱 🙀 Big news from OpenAI! 🙀 🤖 The API Playground now has a Chat mode! 🤖 😻 thanks to @nadipity @_chestercho 😻 | 17.1× | 128k | 526 | 54 | 103 |
| 34 | Dec 13, '24 | you can just do things, like invite @realDonaldTrump to your wedding | 15× | 112k | 1.2k | 13 | 116 |
| 35 | Aug 12, '24 | grtt | 14.8× | 110k | 354 | 10 | 36 |
| 36 | Feb 26, '25 | i bet the slack outage was caused by an ex-openai employee that hasn't fully absorbed slack's culture | 14.3× | 107k | 2.1k | 55 | 65 |
| 37 | Feb 13, '25 | 💡with this update, you can now start asking "gpt-6 wen" | 14× | 105k | 814 | 8 | 28 |
| 38 | Dec 8, '23 | the google doc broke so people texted each other at 2-2:30 am begging people with write access to type their name | 13× | 97k | 233 | 10 | 36 |
| 39 | May 2, '25 | we missed the mark, learned a lot, and wrote it up: | 12.9× | 96k | 604 | 21 | 83 |
| 40 | Mar 7, '25 | View post on X | 12.8× | 96k | 939 | 17 | 71 |
| 41 | Nov 20, '23 | openai is nothing without its people | 12.8× | 96k | 933 | 52 | 15 |
| 42 | Dec 4, '24 | iykyk that it’s really hard to ship or even announce one thing, let alone 12 consecutively lots of things can go wrong, so maybe on those days you’ll see @sama or @kevinweil come o | 12.4× | 93k | 550 | 13 | 47 |
| 43 | Feb 23, '25 | @ibab if i were you i wouldn't throw anyone under the bus so publicly like this & instead run a blameless retro on preventing rogue system message changes going forward ofc do | 12.4× | 93k | 2.3k | 15 | 56 |
| 44 | Dec 27, '23 | out of all the courses i took in college, philosophy still remains the most relevant to my day-to-day job as both (1) a product manager and (2) a pm in AI. so take more philosophy | 12.2× | 91k | 275 | 6 | 68 |
| 45 | Mar 19, '25 | 📢 we’re hiring a PM for model behavior at openai! this role is as fulfilling as it gets for someone thoughtful, curious, and genuine who cares deeply about people & the future — b | 11.4× | 85k | 401 | 20 | 141 |
| 46 | Nov 24, '23 | .@mcleavey is goated | 11× | 82k | 178 | 6 | 40 |
| 47 | Feb 17, '24 | sora for unlimited blackhead extraction videos | 10.6× | 79k | 122 | 0 | 4 |
| 48 | Dec 19, '23 | all my work friends on pto: | 10.4× | 78k | 313 | 6 | 12 |
| 49 | Dec 20, '23 | 🧼 you can now clear your sidebar while still keeping chats! shout-out to @Javi, @AntonTananaev @globetro, @prasad1chakka, @ikeadrift for this quality-of-life improvement while als | 10× | 75k | 138 | 5 | 8 |
| 50 | May 8, '24 | 📖 we just shared the model spec, i.e. a “spec” for openai’s models. it’s a work in progress that we’re sharing for early feedback. it also features profanity & cats, flat earth th | 9.7× | 73k | 246 | 39 | 60 |
| 51 | Apr 29, '24 | memory won't make a difference in your ChatGPT experience on day 1. but! we hope you'll find that the more you use it, the more helpful it gets over time. we're still iterating on | 9.7× | 73k | 236 | 18 | 54 |
| 52 | Aug 9, '24 | "the universe does not want you to ship, but you must do it anyway" | 9.2× | 69k | 242 | 9 | 46 |
| 53 | Aug 31, '22 | DALL·E's canvas just got bigger! Special s/o to @_dschnurr for his end-to-end wizardry -- prototyping, building, making beautiful art, doing press interviews, and so much more! 🪄 | 8.3× | 0 | 894 | 91 | 120 |
| 54 | Sep 30, '25 | ok here's my actual day in the life | 8.2× | 61k | 247 | 9 | 37 |
| 55 | Dec 12, '23 | i still think about this at least 3x a day | 8.2× | 61k | 297 | 12 | 64 |
| 56 | Jan 31, '25 | excited for you to meet o3-mega-high someday 🍃 | 7.8× | 58k | 494 | 16 | 22 |
| 57 | Mar 5, '25 | I shared the following note with the OpenAI team today: | 7.2× | 54k | 251 | 15 | 26 |
| 58 | Jun 5, '25 | full text / blog form: | 7.0× | 52k | 248 | 19 | 93 |
| 59 | Sep 25, '23 | ✨ chatgpt now has a voice – five, actually! hope you'll find one that speaks to you. 💕 | 6.8× | 51k | 208 | 14 | 11 |
| 60 | Jul 30, '25 | if you call everyone engineers meta can't take your researchers 🫨 | 6.8× | 51k | 637 | 20 | 34 |
| 61 | Mar 9, '24 | miwa muwati | 6.4× | 48k | 235 | 4 | 14 |
| 62 | Jan 6, '24 | @rememberlenny @OpenAI "so what should we ship today" – lenny on a random saturday at the office you've been a dream to work with and so core to our shipping culture, maximizing sc | 6.2× | 46k | 186 | 1 | 20 |
| 63 | Sep 24, '24 | 🎙️ designing model behavior for a new interaction paradigm has been more challenging than we expected — some questions we ran into: - can the model come off as flirty or cringey w | 6.2× | 46k | 226 | 17 | 58 |
| 64 | Nov 21, '24 | the moat is shipping through a coup happy 1-year 🎉 | 6.0× | 45k | 241 | 4 | 20 |
| 65 | Oct 26, '24 | our couch was too comfortable, so fiancé and i replaced it with 2 tables to optimize for research focus | 5.7× | 43k | 238 | 1 | 36 |
| 66 | Feb 22, '24 | about yesterday's incident of chatgpt going off the rails: | 5.5× | 41k | 220 | 20 | 30 |
| 67 | Oct 13, '25 | @ sora team, please ban my husband | 5.3× | 40k | 404 | 6 | 42 |
| 68 | May 13, '24 | 3.5 -> 4 -> 40 | 5.0× | 37k | 325 | 16 | 11 |
| 69 | Feb 23, '25 | btw if employees at *any other US-based frontier lab* were told to change the system prompt to tend to the CEO's reputation, they would laugh and simply... not so this is a really | 4.9× | 37k | 399 | 16 | 23 |
| 70 | Aug 3, '24 | oh and you might get to work w @tszzl | 4.8× | 36k | 142 | 2 | 12 |
| 71 | May 8, '25 | 💦💦 | 4.8× | 36k | 194 | 3 | 21 |
| 72 | Dec 3, '22 | ChatGPT can draw better than DALL-E: | 4.8× | 0 | 556 | 54 | 37 |
| 73 | Oct 1, '24 | if you're at devday today, please grab me (or any random employee) and tell us all about your model behavior peeves from a developer perspective! ... or just reply here too | 4.7× | 35k | 287 | 6 | 32 |
| 74 | Aug 7, '25 | 🤖 introducing _robot_ with gpt-5! it doesn't do small talk. it doesn't ask you how you're doing. it doesn't believe in humor... although i personally find it pretty funny. 🧵 this | 4.6× | 34k | 156 | 8 | 35 |
| 75 | Mar 27, '25 | full text / blog form: | 4.5× | 34k | 152 | 5 | 54 |
| 76 | Apr 1, '25 | my sidebar rn: | 4.2× | 31k | 343 | 12 | 3 |
| 77 | Apr 2, '25 | gm | 4.2× | 31k | 117 | 2 | 3 |
| 78 | Mar 5, '25 | go try 4.5! it was crafted with a lot of love. | 4.0× | 30k | 255 | 11 | 11 |
| 79 | Sep 20, '24 | he's a core contributor there though, not a foundational contributor | 3.9× | 30k | 121 | 0 | 12 |
| 80 | Apr 1, '25 | the masochist in you will keep this as your default like i did | 3.9× | 29k | 192 | 4 | 18 |
| 81 | May 13, '24 | 💕 | 3.9× | 29k | 125 | 2 | 7 |
| 82 | Nov 22, '23 | @tszzl should we ship the thing next week | 3.7× | 28k | 321 | 8 | 8 |
| 83 | Sep 12, '24 | o1 is adorable when it overthinks things: (not cute in an anthropomorphic sense, cute like a lost roomba!) | 3.7× | 27k | 220 | 11 | 33 |
| 84 | Oct 12, '25 | the opposite of a servant leader is a dominatrix peon | 3.6× | 27k | 135 | 8 | 16 |
| 85 | Feb 12, '25 | 📖 model spec v2! it's the latest version of the doc that outlines our desired intended behavior for openai's models. some of additions shouldn't be controversial, but some are def | 3.5× | 26k | 136 | 12 | 17 |
| 86 | Nov 30, '23 | Happy birthday @ChatGPTapp 🎂 | ⚡ 3.3× | 475k | 376 | 33 | 11 |
| 87 | Feb 28, '24 | "what do you do after work" | 3.0× | 23k | 174 | 4 | 10 |
| 88 | Jun 5, '25 | PSA the further out your wedding, the better models will be at vow writing | 3.0× | 22k | 167 | 3 | 16 |
| 89 | Jul 6, '25 | i've had _insane_ luck multiple times career-wise, one of which was getting to learn directly from @pxd, at openai's pace i recommend listening to the full podcast -- @lennysan pul | 2.9× | 21k | 163 | 6 | 107 |
| 90 | Jul 10, '25 | this is how @tszzl talks in real life btw | 2.9× | 21k | 243 | 3 | 9 |
| 91 | Mar 31, '25 | this means millions of people are realizing today how powerful & delightful AI can be <3 | 2.7× | 20k | 195 | 4 | 5 |
| 92 | Nov 21, '23 | s/o to @AlecRad @_jongwook_kim @txhf and @mcleavey – the research team behind voice – for the magic here. go try it out for yourself!! <3 | ⚡ 2.6× | 483k | 255 | 25 | 42 |
| 93 | May 13, '24 | fave part of all this is @markchen90 and @barret_zoph asking for math help bc they def need that | 2.6× | 19k | 134 | 4 | 6 |
| 94 | Mar 28, '25 | also: a lot of imagegen refusals are due to the model hallucinating policies that don’t exist, amidst this shift. bear with us as we try to get the model to follow the policy & | 2.4× | 18k | 127 | 3 | 14 |
| 95 | Aug 29, '25 | (oh can journalists confirm if i can shitpost more now without it being tied to oai) | 1.9× | 14k | 116 | 1 | 5 |
| 96 | Aug 4, '25 | 📈📈📈 | 1.8× | 14k | 146 | 3 | 5 |
| 97 | Dec 12, '24 | shout-out to Members of Christmas Staff 🦌🛷🎅🏻 (this may be my favorite launch of the year, excited to see everyone interact with it!!) | 1.6× | 12k | 134 | 2 | 6 |
| 98 | May 18, '22 | I don’t usually “shill” where I work, but OpenAI is a truly special place comprised of brilliant people, focus, and optimism — and I’m so lucky to be here, right now. | 1.4× | 0 | 170 | 9 | 17 |
| 99 | Jul 13, '22 | At some point I'd love to write about all the insanely hard problems + brilliant solutions our team came up with get to this point and beyond. Will just brag for now: currently liv | 1.3× | 0 | 170 | 5 | 8 |
| 100 | May 18, '22 | We're ramping up DALL·E invites to 1K per week! | 1.0× | 0 | 127 | 6 | 2 |
| 101 | Feb 26, '25 | @asimdotshrestha worker 17 is also an ex-openai employee that hasn't fully absorbed sweatshop culture | 0.7× | 5.5k | 169 | 3 | 3 |
@karpathy
@levelsio
@paulg
@naval
@marclou
@illyism