Kevin Weil

Could not agree more with @gdb.

This is a great take. Human intelligence and computer intelligence are not different points on a line, but in a high dimensional space—and we should explore more of the space.

Five Erdos problems at once! The proofs are getting more elegant as the models improve 👀

Btw: @hemal built this as a simple skill in a few hours, since Prism is powered by Codex. More to come. And if you love the idea of building AI-native products to accelerate science, we're hiring a lead designer for Prism: apply here! https://jobs.ashbyhq.com/openai/7e265aee-6842-4fc1-bf28-1df3ea4ee013

💥 New in Prism today: Paper Review, an AI workflow for reviewing technical and scientific papers. This is the opposite of AI slop: we're using AI to improve scientific rigor, correctness, and reproducibility.

RT Mehtaab Sawhney In a beautiful recent paper, Vishesh Jain and Clayton Mizgerd used GPT-5.4 Pro to prove a striking result in the theory of Markov chains: https://arxiv.org/pdf/2604.03937 They study the adjacent transposition Markov chain on the symmetric group. A conjecture of Fill, recently settled by Greaves and Zhu, determined which parameters of this chain maximize the spectral gap, a natural quantity controlling how fast the chain mixes. Jain and Mizgerd go further and characterize exactly when this extremal spectral gap is achieved, answering another question of Fill. As they explain in the paper, once the first part was in place, GPT-5.4 Pro was able to one-shot generate the second part of the main result. From talking with the authors, my understanding is that this would likely have taken substantial effort without GPT. Furthermore even given the first part, several ingredients, such as the piecewise eigenvector construction in Proposition 6.6, were new to them. Just another example of how AI is already changing the everyday practice of research mathematics. Original tweet: https://x.com/mehtaab_sawhney/status/2041354267286737243

RT Simon Smith I’ve been critical of OpenAI lately, but for the past three weeks my family has been dealing with a health issue with my dad, and a ChatGPT shared project with live document syncing has been essential to organizing and understanding everything happening. Me, my four siblings, my mom, and my dad have faced an onslaught of information from various doctors and nurses, which we’ve captured in hundreds of text messages and documents and scans and you name it. ChatGPT has helped us collect this information in a single place, make sense of it, and interrogate it to make the most informed decisions possible. Also, credit where due: Claude played an important role as well, by ingesting iMessages and synthesizing summarizes from them to upload to ChatGPT, as well as by extracting text from a bunch of HEIC document scans. I think those of us, like me, excited at AI’s potential get frustrated when we can see issues so clearly, like ChatGPT’s bad design skills, and Claude’s increasing instability and confusing usage consumption. But at times like this I’m reminded of how incredible this technology already is, letting me and my family make sense and act on hundreds of pieces of information, empowering us in the face of a disjointed and fragmented healthcare system. Original tweet: https://x.com/_simonsmith/status/2040539824034115676

Not only is AI solving more open problems—its proofs are getting more elegant as the models improve

RT roon god I love technology Original tweet: https://x.com/tszzl/status/2038856932371681493

RT Chris Hadfield If schedule holds, these 3 giant rockets will launch in the next 3 weeks. From left to right: New Glenn - satellite launch now, planned for the Moon Starship - test flight 12 now, planned for the Moon Artemis - to the Moon and back with 4 crew aboard Pushing the very edge of our capability as we learn how to more safely & cheaply reach space, to explore all that exists beyond. @nasa @SpaceX @blueorigin Original tweet: https://x.com/Cmdr_Hadfield/status/2038629519012061303

"An open source workflow for producing a personalized mRNA cancer vaccine"

Paul used ChatGPT + AlphaFold to create a personalized mRNA vaccine protocol for his dog's cancer. This is a glimpse of the future, with AI accelerating personalized medicine. We have to make it easier for people to do this for people, not just dogs!

RT Katherine Boyle Great to see this. Gwynne Shotwell should be a household name. One of the most impressive business leaders of our time. Original tweet: https://x.com/KTmBoyle/status/2037141887363043838

👀

RT Leeham GPT-5.4 Pro fully resolves a Machine Learning adjacent Open Math problem from http://solveall.org! I submitted the solution a few weeks ago and just checking now, it has been accepted. This marks the 2nd problem on the site to be fully resolved! https://solveall.org/problem/gaussian-correlation-inequality-extensions Original tweet: https://x.com/Liam06972452/status/2036266860417540425

I’m inspired by young people putting frontier AI to work in science in new and surprising ways. We recently met a young astronomer who used AI to analyze massive datasets. He identified 1.5 million new celestial objects in space, and the head of NASA offered him a job and a ride in a fighter jet as a signing bonus! (https://x.com/rookisaacman/status/2004772750494499104) To celebrate stand-out students and recent grads like this—and help keep their ideas moving—we’re launching a new program: ChatGPT 26. If you're selected, we'll host you at OpenAI HQ in SF, give you access to our latest tech, and give you a $10K cash grant to keep pushing your ideas forward. If this sounds like you, we’d love to meet you and hear your story. Apply or nominate someone here: http://chatgpt26.com

Technology is so rad. @neuralink is amazing.

RT Dwarkesh Patel When Copernicus proposed heliocentrism in 1543, it was actually less accurate than Ptolemy's geocentric model - a system refined over 1,400 years with epicycles precisely tuned to match observed planetary positions. It took another 70 years before Kepler, working from Tycho Brahe's unprecedentedly precise observations, replaced Copernicus’s circles with ellipses - finally making heliocentrism empirically superior. Terence Tao's point is that science needs a high temperature setting. If we only fund and follow what's most state of the art today, we kill the ideas that might need decades of work to surpass some overall plateau. Original tweet: https://x.com/dwarkesh_sp/status/2035114158241587221

I'll put a lot of money on the over, @jpatel41 :)

RT Fidji Simo This news came out a little earlier than we planned; we're excited to be building a deployment arm and will share more details soon. Companies have a ton of urgency to deploy AI in their organizations and we’re sprinting to meet that demand. More than 1 million businesses run on OpenAI products. Codex is now at 2M+ weekly active users, up nearly 4x since the start of the year. API usage jumped 20% in the week after GPT-5.4 launched. And Frontier, which launched last month to help enterprises build, deploy, and manage AI coworkers that can do real work, has way more demand than we can handle. That's why we launched Frontier Alliances so we leverage our ecosystem of partners to scale. And that is also why we are launching a dedicated deployment arm tasked with embedding Forward Deployed Engineers deeply inside of enterprises.  This project has been in the works with our investor and alliance partners since last December, and we are grateful for them and their partnership. We’re still early, but the speed of adoption is a clear signal of where this is headed. We're excited to not just be building these technologies but also building many ways for companies to deploy them and get impact. https://www.reuters.com/business/openai-courts-private-equity-join-enterprise-ai-venture-sources-say-2026-03-16/ Original tweet: https://x.com/fidjissimo/status/2033537381907710092

This is the right take.

This is beautiful, and true

RT martin_casado 5.4 is really special. Original tweet: https://x.com/martin_casado/status/2031536708706250936

So excited to work together! I have a feeling it's going to be a productive summer :)

If true, this would be the first of @EpochAIResearch's Frontier Math open problems to be resolved by AI. "The result emerged from a single GPT-5.4 Pro run and was subsequently refined into Lean with GPT-5.4 XHigh which ran for a few hours."

A look at the future/present

This is a pretty mind-blowing look inside GPT-4 and its reasoning abilities. Definitely worth your team to read.

Codex for Open Source is an awesome idea. OSS maintainers get API credits, 6 months of ChatGPT Pro with Codex, and access to Codex Security as needed.

RT Daniel McAuley some psychopath on the internal codex leaderboard hit 100B tokens in the last week Original tweet: https://x.com/_dmca/status/2029810231325380725

💥 GPT 5.4 is launching today! It's our best model ever, and it's also the most capable scientific model we've ever released. GPT 5.4 Pro in particular is 🤯 based on early testing with scientists and mathematicians.

We integrated the Codex harness into Prism (http://prism.openai.com) — this means you get skills, reasoning levels, and the raw tenacity of the Codex model in your LaTeX environment. Oh and we also built version mgmt into Prism, which was one of the top requests. See below thread from @vicapow for some great examples of the power of Codex inside Prism 👇

💥 AI accelerating high energy physics Just a few weeks after the gluon scattering paper, this morning we posted the more complicated graviton scattering analogue. See below for more from @ALupsasca 👇

RT Mehtaab Sawhney Last week in a beautiful preprint, Dmitrii Zakharov proved a bound for a two-family version of a word-overlap problem, but the first version left a constant factor gap of e. That gap was quickly improved by a mix of GPT-5.2 Pro and human collaborators. Even more recently, an internal OpenAI model obtained the asymptotically sharp bound. Dmitrii incorporated the argument into v2 of his paper https://arxiv.org/pdf/2602.20143; see his AI statement for more details. Original tweet: https://x.com/mehtaab_sawhney/status/2029052169727754249

RT Joshua Achiam A firm commitment to the principle that AGI companies have to devolve power to democracies and avoid unduly concentrating power in themselves, even when that leads to uncomfortable places, is something I will not regret. Original tweet: https://x.com/jachiam0/status/2027866890278998126

🇺🇸

RT Sam Altman We have raised a $110 billion round of funding from Amazon, NVIDIA, and SoftBank. We are grateful for the support from our partners, and have a lot of work to do to bring you the tools you deserve. Original tweet: https://x.com/sama/status/2027386252555919386

So excited for the new world where everyone can build

RT Mehtaab Sawhney We just posted a paper solving Erdos #846, which was solved by an internal model at OpenAI (https://cdn.openai.com/infinite-sets/main_single_clean3.pdf). While the problem can also be derived from an earlier paper in the literature, the proof by the internal model was one of the first instances where I smiled reading the proof. Original tweet: https://x.com/mehtaab_sawhney/status/2026716221933343147

Instantly one of the great sports photos of all time.

Daniel has a really interesting perspective on the future of Math and AI—well worth a read 👇

RT Greg Brockman seeing so much positive progress across each part of openai right now, very proud of the team Original tweet: https://x.com/gdb/status/2024985187579560366

RT Tejal Patwardhan new in Nature about our wet lab evals work! Original tweet: https://x.com/tejalpatwardhan/status/2024636639126102513

More on the gluon scattering/GPT 5.2 paper from @ALupsasca below 👇 If you're in the Boston area on Tuesday, go see his lecture at Harvard!

RT Peter Steinberger 🦞 I'm joining @OpenAI to bring agents to everyone. @OpenClaw is becoming a foundation: open, independent, and just getting started.🦞 https://steipete.me/posts/2026/openclaw Original tweet: https://x.com/steipete/status/2023154018714100102

RT Sam Altman We went from AI systems that struggled to do grade school math to AI systems that can solve research-level math problems in just a few years. I agree with Jakub this is perhaps the most important eval now. I am also pretty sure the main reaction will be "it's not that hard" :) Original tweet: https://x.com/sama/status/2022729068949717182

AI 🤝 Mathematics. The First Proof challenge was a great idea and I'm looking forward to seeing the results! Great color on OpenAI's submission from @merettm below 👇

Not gonna lie it's kinda amazing to see gluon scattering trending on Twitter... my worlds colliding https://x.com/i/trending/2022395629113377198

👇💥

Coding at 1000 tokens/sec is a mind-expanding experience. You have to try this.

RT Derya Unutmaz, MD As I mentioned, I started a really crazy project with the new OpenAI GPT-5.2 Deep Research; I created a 110-chapter textbook focused only on T cells, the immune cells I’ve studied for 35 years, & it’s over 1,000 pages long! Sharing the first 15 chapters here: https://tcell-textbook.netlify.app/ I’ll assemble and share the rest soon. Next, I plan to create similar, even more ambitious textbooks on the entire immune system, cancer, aging, and ME/CFS. Because now you can just do things! Original tweet: https://x.com/DeryaTR_/status/2021780855568417244

Could not agree more. Curious for everyone's take on how we best address this, given the relative speed of model improvement vs peer review.

Can confirm

RT Michael Baym A lot of people talking about this like it's a bunch of vibecoders who hooked up an LLM to some robots In reality Joy is one of the best experimentalists I've ever met (before she went to OpenAI), and Ginkgo has spent nearly two decades developing best-in-class lab automation Original tweet: https://x.com/baym/status/2019799194400145568

Coding models are amazing, but "GPT-5 lowers the cost of cell-free protein synthesis by 40%" is also pretty rad. This was all done in a robotic lab in partnership with @Ginkgo. AI + Science!

So many AI graphs are in log scale, you forget how wild it really is.

🤣🤣🤣

AI doesn't just accelerate science in big ways, it accelerates it in a thousand incremental compounding ways too.

RT Noam Brown Every ~6 months I hear people claim @OpenAI isn’t doing real research and is just incrementally improving ChatGPT. I even heard it right before 🍓/o1. In my opinion @OpenAI is the best frontier lab to do research at today. Building an AI research intern in 2026 is not hype. Original tweet: https://x.com/polynoamial/status/2018792698107634108

If you're a scientist and/or write in LaTeX, try Prism at http://prism.openai.com and let us know what you think! Taking feedback and feature requests here and turning them into code daily with Codex.

I've gotten to the point where if I'm sitting in a meeting and I don't have a Codex prompt running, I feel anxious for "wasting" an hour I could have been making parallel progress. It's happened super quickly and totally changed how I think.

If you're interested in the future of AI and science/mathematics, I highly recommend watching @SebastienBubeck's talk "Recent Advances in LLMs for Mathematics." Seb is an amazing AI researcher at OpenAI and former math professor at Princeton. https://www.youtube.com/watch?v=MH3lG7V7SuU

Thanks for having me on, @jordihays @johncoogan! Always super fun to talk with you on TBPN.

Welcome Mehtaab—so excited to get to work with you!

It seems some people are misinterpreting comments @thefriley made in Davos. To be 100% clear: she was not saying that OpenAI plans to take a share of individual users’, entrepreneurs’, or scientists' discoveries. We’ve heard interest from some large organizations in licensing or IP-based partnerships, and we’re open to exploring creative ways to partner and align incentives. That’s not something we’re doing today, and if we do it in the future, it would be a bespoke agreement with a company, not something that would impact individual users. As it relates to Prism: you log into Prism with your ChatGPT account. In ChatGPT, you control in settings whether your (anonymized) data is used to improve our models. Prism follows the setting you've chosen in ChatGPT. So you're in control. https://x.com/kevinweil/status/2016210486778642808

💥 Today we’re introducing Prism—a free, AI-native workspace for scientists to write and collaborate on research, powered by GPT-5.2. Accelerating science requires progress on two fronts: 1. Frontier AI models that use scientific tools and can tackle the hardest problems 2. Integrating that AI into the products scientists use every day Prism is free to anyone with a ChatGPT account, with unlimited projects and collaborators. Try it today at http://prism.openai.com—would love to hear your feedback.

👀💥 5️⃣.2️⃣

Congrats @JohnnotJon! Granted I'm 42, but I still think you (and Airpost) are cool.

Love this prompt from @joannejang! Here's mine. What's yours? Asking it to explain it afterwards is fun too.

Another open Erdös problem solved by GPT 5.2, confirmed by Terence Tao. Great work @neelsomani.

RT Tibo Codex is having a fantastic start to 2026, we're adding compute at unprecedented pace and OpenAI itself is going through a fast evolution phase as we co-evolve how we build together with Codex's capabilities week after week. It's going to be a fun year. Original tweet: https://x.com/thsottiaux/status/2012416863356002373

GPT 5.2 ran uninterrupted for *one week* and wrote *3 million* lines of code. The future is going to be awesome

Wow, the third Erdos problem solved via GPT 5.2 this week!

2026 is going to be an exciting year for science! AI will be a meaningful accelerant and it couldn't come at a better time.

RT dave kasten Fascinating: Maryland becomes first state govt to try to ship a textfile to help LLMs navigate government services (llms.txt) Original tweet: https://x.com/David_Kasten/status/2009305237949931550

RT Fidji Simo The launch of ChatGPT Health is really personal for me. I know how hard it can be to navigate the healthcare system (even with great care). AI can help patients and doctors with some of the biggest issues. More here: https://fidjisimo.substack.com/p/chatgpt-health Original tweet: https://x.com/fidjissimo/status/2008978500557131893

Super cool! GPT-5.2 🤝 @HarmonicMath to solve and formalize Erdos #728.

Great advice for 2026 that would have made zero sense in 2025

RT Greg Brockman two big themes of AI in 2026 will be enterprise agent adoption and scientific acceleration Original tweet: https://x.com/gdb/status/2006584251521839141

Congrats to the @OpenAI research team—GPT 5.2 is an incredible model!

This is wild and impressive... and will seem commonplace in 12-24 months.

High school student uses AI to discover 1M+ objects humans missed in astronomical data. Head of NASA openly recruiting him through Twitter with a fighter jet ride included. All my worlds colliding. I love everything about this.

RT Aaron Levie http://x.com/i/article/2004648738762227713 Original tweet: https://x.com/levie/status/2004654686629163154

I just straight up lost to my 11yo in chess, AMA

RT Poetiq We finally had a moment to run our system with GPT-5.2 X-High on ARC-AGI-2! Using the same Poetiq harness as before, we saw results as high as 75% at under $8 / problem using GPT-5.2 X-High on the full PUBLIC-EVAL dataset. This beats the previous SOTA by ~15 percentage points. Original tweet: https://x.com/poetiq_ai/status/2003546910427361402

This is great @ivanhzhao!

RT Alex Predhome, Track and Field Enjoyer Time for a Christmas classic Original tweet: https://x.com/Predamame/status/2002908392558678344

RT roon the primary criticism of AI you hear has nothing to do with water use or existential risk whatsoever: most people just think it’s fake and doesn’t work and is a tremendous bubble eating intellectual property while emitting useless slop along the way. when GPT-5 came out and perhaps didn’t live up to what people were expecting for a full version bump, the timeline reaction was not mild, it was a full-scale meltdown. there are many intelligent (and unintelligent) people who latched onto this moment to declare AI scaling over, thousands of viral tweets, still a prevailing view in many circles. The financial-cultural phenomenon of machine intelligence is one of the most powerful in decades, and there are a lot of people who would like for its position to be weakened, many outright celebrating its losses and setback. Michael burry of ‘Big Short’ fame, unfortunately the type of guy to predict 12 of the last 3 recessions, has bet himself into insolvency on the AI bubble’s collapse one of the stranger things about this time is that there are very few secrets, and very little reason to be so misinformed. model labs have very little space in between creating new capabilities and launching them to the public. The view among the well informed public and not just “lab insiders” is that machine intelligence is absurdly joyfully smart at so many new things every month. It’s actively contributing on the cutting edge of programming and math and science. Sebastian Bubeck and co’s recent paper reports that GPT5-pro is capable of producing results on the frontier of theoretical physics research, Terry Tao wrote a blog about “vibe-proving” Erdos problems with the auto-formalization AI Aristotle. You can read that these scientists are using it to actively contribute to black hole physics, tighten mathematical bounds in optimization theory, churning morasses of biomedical data into real insight. Google Deepmind, from the way they are signalling, seems to be slowly closing a dragn...

The Genesis Mission is a brilliant set of ideas. Very excited to deepen @OpenAI's partnership with the DoE and the National Labs in the name of AI and national security 🇺🇸

💀

This is super exciting 👀

💥 The new ChatGPT imagegen is here! Plus there's a super fun Images section in the ChatGPT app now too. Built-in prompts make it easy to gen great images. Try it and share what you make below!

Science 🤝 GPT-5. Our new FrontierScience benchmark will be a valuable way to measure the performance of AI models on hard chemistry, biology, physics, and more. Plus, GPT-5 operating in a wet lab environment suggested experiments to increase a molecular cloning protocol's efficiency by 79x. Great thread below 👇

RT Daniel Litt OK, I think GPT 5.2 Pro is actually a step change in usefulness for my applications (algebraic geometry/number theory research). Original tweet: https://x.com/littmath/status/2000636724574302478

RT Sam Altman GPT-5.2 exceeded a trillion tokens in the API on its first day of availability and is growing fast! Original tweet: https://x.com/sama/status/1999624463013544024

Love this from @Opendoor 🇺🇸

RT Sebastien Bubeck btw the paper described in this thread is on arxiv already: https://arxiv.org/abs/2512.10220 ! Original tweet: https://x.com/SebastienBubeck/status/1999540978676355478

RT Marc Andreessen 🇺🇸 It’s time to build. Original tweet: https://x.com/pmarca/status/1997109742200934620

Want a masterclass in using ChatGPT? Read this account. I work here and helped build these products, and it still blew my mind.

RT Brad Gerstner 🇺🇸🚀 Original tweet: https://x.com/altcap/status/1995860986407096568

💯. We want more experienced industry leaders like @DavidSacks in government, especially in critical and highly technical areas like AI.