Akila Welihinda

@ypatil125 @appliedcompute woohoo congrats @ypatil125!

Today we’re announcing $80M in new financing for @appliedcompute, to help enterprises go from general to specific intelligence. Our view is that the winners of this era will be the companies that turn their information, and expertise into intelligent systems that compound over time. They’ll have agents that get sharper every week by learning from real decisions, outcomes, and feedback.

@tszzl you’re absolutely right

feeling sort of gullible today, maybe due to selection effects

@brendangregg welcome!!

New blog post: Why I joined OpenAI brendangregg.com/blog/2026-02-0…

@barret_zoph Welcome back Barret! 😄

Super excited to join the team!

@thsottiaux 5.2 codex xhigh feels like a massive step up. it’s faster than 5.2 and smarter. Even when I’m on PTO i always have one session running

GPT-5.2-Codex is now by and far the most popular model in Codex. Production distribution shows significant token-efficiency gain at equal reasoning effort compared to GPT-5.2 in Codex. Will share plots soon. What's your experience?

@jiangelaa @worktrace_ai @deepakv91 congrats @jiangelaa!

👋@worktrace_ai is out of stealth! Which also means I've officially rejoined the workforce...I couldn't help but join @deepakv91 to pursue this vision together. I really think we & our amazing team are on track to make a meaningful difference in bridging the AI divide. Join us!

@sama completely agree. codex has been killing it 🚀

It has been amazing to watch the progress of the Codex team; they are beasts. The product/model is already so good and will get much better; I believe they will create the best and most important product in the space, and enable so much downstream work.

@tszzl fwiw I also had the same issue

everyone got mad at me for this one but i think it’s fair. i tried the app and it didnt have gemini 3 at the time. I thought maybe i need to upgrade my plan, which has two diff buttons that led to confusing flows that went nowhere

@uninsightful I feel like Waymo’s are increasingly running this city, especially at night. After sunset, I see more Waymo’s than any other vehicle

every time I come off a week of taking waymos in SF: 1. it feels increasingly strange to return to a non-autonomous city (just as it felt weird to be in cities that didn't have uber yet in 2014-2016) 2. I come away feeling like we continue to under-discuss the second order effects of self-driving inevitability + ubiquity I think the indifference in the air is largely a function of how gradual (relatively) the rollout of AVs has been and will continue to be

@appliedcompute congrats @ypatil125! we miss ya over here

Generalists are useful, but it’s not enough to be smart. Advances come from specialists, whether human or machine. To have an edge, agents need specific expertise, within specific companies, built on models trained on specific data. We call this Specific Intelligence. It's what we're building at Applied Compute. We unlock the latent knowledge inside a company, use it to train custom models, and deploy an in-house agent workforce that reports to your team. We work with sophisticated companies that have already captured early gains from general models, like @cognition, @DoorDash, and @mercor_ai. They’re pulling even further ahead with proprietary in-house agents that don’t need to wait for the next public model release. Together, we are building and validating models and agents in days instead of months, achieving state-of-the-art performance on customer evals. Our team has high density and low latency. Our founders all worked on different parts of this problem while they were researchers at OpenAI — @ypatil125 as a key member on the agentic software engineer effort (Codex), @rhythmrg as a core contributor to the first RL-trained reasoning model (o1), and @lindensli as a core contributor on ML systems and infrastructure for RL training. Two-thirds of the team are former founders, and everyone brings a deep technical background, from top AI researchers to Math Olympiad winners. We are backed by $80M in funding from Benchmark, Sequoia, Lux, Elad Gil, Victor Lazarte, Omri Casspi, and others. With their support, we are growing the team, scaling deployments, and bringing to market the first generation of agent workforces built on specific models. In short: 1. We are building Specific Intelligence for specific work at specific companies. 2. That will power in-house agent workforces to support their human bosses. 3. That in turn will unlock AI’s full potential through humanity’s greatest engine of progress: thriving corporations in a free market.

@SemiAnalysis_ @dylan522p @dwarkesh_sp @dwarkeshpodcast what remix is this?

POV: when the cousin of @dylan522p , @dwarkesh_sp reaches 1mil subscribers on his @dwarkeshpodcast channel after his banger interview drops

@tszzl Honestly same

i consider leaving all the time, going to a real city, having some new adventures. but i have been cursed with an abounding and obsessive love of technology

@gdb silicon working when I’m sleeping is the best feeling

that feeling when you've finished debugging, kicked off your experiments, and get to relax waiting for the results to roll in

@abshkbh Glad you joined our team!

For the past year I’ve been building Arrakis on a single thesis: with the right tools and secure environments, LLMs can reliably do complex work. This journey started two years ago when I left a stable role at Google to work on early coding agents. While still at Google, I wrote a long email to @gdb about how a systems engineer could break into AI. Arrakis opened doors and has led to a full-circle moment: I’ve joined @OpenAI to work on Agent Infrastructure in the Scaling org. It’s a privilege to help people through smarter models and agents. I’m especially excited about our coding initiatives. Thank you @gdb and @paulashbourne for the opportunity. Looking back, the biggest risk was not taking one!

@thsottiaux I moved to 5-codex on Monday. Haven’t looked back since

After trying gpt-5-codex, which one will you keep as your daily driver for coding?

@MostafaRohani @wyndwarrior @robin_a_brown @bminaiev @_lorenzkuhn @clavera_i @andresnds @ahelkky amazing job! i personally find this milestone the most exciting because i used to dabble in ICPC

8/n Research is very much a team sport, and it was so much fun to work with @wyndwarrior, @robin_a_brown, and @bminaiev on prepping for the competition—and with @_lorenzkuhn, @clavera_i, @andresnds, and @ahelkky on competitive programming in general. I remember when I joined OpenAI, our best AI models could barely solve Codeforces easy problems. It’s incredible to see the rate of progress over these last two years. And with this team, I’m really excited to see what we achieve in the next two!

1/n I’m really excited to share that our @OpenAI reasoning system got a perfect score of 12/12 during the 2025 ICPC World Finals, the premier collegiate programming competition where top university teams from around the world solve complex algorithmic problems. This would have placed it first among all human participants. 🥇🥇

@McaleerStephen Oh man! Gonna miss our chats about coding agents and best of luck in your next journey

Today was my last day at OpenAI. It's tough to leave given all the great safety work going on but I decided to deepen my focus on scalable oversight and frontier AI safety. Extremely grateful for my time at such a legendary company!

@TheRealRPuri @OpenAI Definitely gonna miss ya @TheRealRPuri

Yesterday was my last day at @OpenAI. Feeling grateful for everything I’ve learned and the incredible people I got to work with. I’m sharing my resignation letter here for colleagues and friends across the industry: "To my dear friends and colleagues, I wanted to share that I’ll be leaving OpenAI, with 9/3 as my last day. Don’t worry—I’m not heading to any of the existing frontier labs. I’ve chosen a different path for my career, one that reflects how I’ve grown over the years and how I hope to continue growing moving forward. A few weeks ago, I turned 30. It made me pause and realize just how quickly time has passed—how much of my 20s, and really the past decade, I’ve devoted to this field. I joined OpenAI just weeks before GPT-3 launched, and now we’ve released GPT-5. It’s incredible to see how far we’ve come: our AI can now see, speak, and think. I still remember talking with Greg about model scaling and interviewing with Alec, Jongwook, Jakub, and Szymon for the language and reasoning teams five years ago. To see the organization actually accomplish this vision is surreal. It’s been eight years since I scaled my first LLM, and more than half a decade—the majority of my 20s—since joining OpenAI. Many of the ideas I hoped to explore at the beginning, I now feel we’ve realized together. That’s a special feeling. While writing my thank-yous, I was struck by how many of our peers are no longer at OpenAI. Yet they remain part of this mission through our memories and the work we shared. That gave me a deep sense of comfort—that even as people come and go, the mission and the connections we built continue to carry forward. Lately, I’ve been reflecting on connection. It’s been extraordinary to connect our research with the world, and to see our work help people feel more connected to their humanity and to the human experience. I can only hope that continues, and that our work helps people feel more connected with one another—the world needs this now more than ever. Even though I’m leaving, I feel deeply connected to all of you through the memories and experiences we’ve shared—the good, the bad, the awkward, and the sublime. Thank you all for the opportunity to build and create together over these years. I’ll share more about what’s next in due time."

@hTrapVader Does it support Flow though? 😂

Edge is the best browser

@LionelAccount @nextjs It couldn’t use the browser or take screenshots

@awelm_ @nextjs Oh wow Does it open browser and do the changes ?

GPT-5 is cracked at frontend. It fully migrated my blog awelm.com from Hugo to @nextjs with a few prompts. It also added these new features: