Operator
“First impression of operator: 1) pretty neat for the demo use cases (although i’d personally never use it to book flights) 2) Misclicks a lot on buttons, usually by a few pixels, wonder if its a viewport issue 3) The take control is pretty clunky. Really destroys the flow”
Introduction to Operator & Agents – YouTube
“Initial thoughts on Operator: – SOTA OSWorld/WebArena means actual meaningful model advance, not just ui/product wrapper. OAI always excels at this (model+product progress), as we discuss in our @karinanguyen_ episode today – interesting that Anthropic Computer Use is a free” / X
We Tried OpenAI’s New Agent—Here’s What We Found
“A research preview of Operator, an agent that can use its own browser to perform tasks for you.”
Introducing Operator research preview | OpenAI
CUA_eval_extra_information.pdf
This document includes extra information to how we evaluated our Computer Using Agent, including (browser/VM) environments, prompts, sampling parameters, and scoring procedures.
“Here’s how this tweet happened! @OpenAI added a share mechanism to Operator”
“Hats off to the @openai team. The ability to use memory and fill out paperwork with ChatGPT is saving me literally hundreds of hours and has stopped me from making some critical errors in the process of dealing with complex insurance and mortgage negotiations related to our” / X
“I am building smooth operator, an alternative to openai operator that – won’t cost $200/month and – uses a combination of the best models on the market, e.g. R1 and screengrasp link below.”
Stargate
OpenAI on X: “Announcing The Stargate Project The Stargate Project is a new company which intends to invest $500 billion over the next four years building new AI infrastructure for OpenAI in the United States. We will begin deploying $100 billion immediately. This infrastructure will secure”
“Wow! Stargate Project will invest $500B over the next 4 years – that’s ~0.4% of US GDP over that period. For comparison, the inflation-adjusted dollars spent on other large undertakings: • Interstate Highway System: ~$650B • Apollo Program: ~$280B • Manhattan Project: ~$35B” / X
“🚨 MASAYOSHI SON: “Mr. President, last month I came to celebrate your winning and promised $100B. And you told me go for $200B. Now I came back with $500B. This is because as you say, this is the beginning of the Golden Age. We wouldn’t have decided this unless you won.” 🔥 🏆”
“Announcing The Stargate Project The Stargate Project is a new company which intends to invest $500 billion over the next four years building new AI infrastructure for OpenAI in the United States. We will begin deploying $100 billion immediately. This infrastructure will secure” / X
“”Now I come back with $500 billion” 😎”
Microsoft and OpenAI Plot $100 Billion Stargate AI Supercomputer — The Information
“Stargate is probably a grift. Not in the sense that $500B will be stolen or anything but just in that it’s a commitment to spend money on capex that these corporations have already earmarked, but now presented as some proactive Trumpist initiative. They’re racing anyway.” / X
Announcing The Stargate Project | OpenAI
“Thank you to President Trump for announcing Stargate Project with us today. $500B for AI data centers for OpenAI, built in the US. 🇺🇸”
“We are living in a timeline where a non-US company is keeping the original mission of OpenAI alive – truly open, frontier research that empowers all. It makes no sense. The most entertaining outcome is the most likely. DeepSeek-R1 not only open-sources a barrage of models but”
OpenAI Product Chief on Trump-Backed ‘Stargate,’ New AI Models and Agents | WSJ – YouTube
“@OpenAI They don’t actually have the money” / X
“@elonmusk @OpenAI wrong, as you surely know. want to come visit the first site already under way? this is great for the country. i realize what is great for the country isn’t always what’s optimal for your companies, but in your new role i hope you’ll mostly put 🇺🇸 first.” / X
Other
Sam Altman on X: “thank you to the external safety researchers who tested o3-mini. we have now finalized a version and are beginning the release process; planning to ship in ~a couple of weeks. also, we heard the feedback: will launch api and chatgpt at the same time! (it’s very good.)”
“leaked benchmark: o3 pro solved problems we thought were 5 years away. sam’s team is trying to figure out how it did it. something unprecedented is happening.” / X
Yuntian Deng on X: “Is OpenAI’s o1 a good calculator? We tested it on up to 20×20 multiplication—o1 solves up to 9×9 multiplication with decent accuracy, while gpt-4o struggles beyond 4×4. For context, this task is solvable by a small LM using implicit CoT with stepwise internalization.”
“o1 pro can really think long and hard, if you ask it the right questions”
(2) OpenAI #10: Reflections – by Zvi Mowshowitz
“In chat or through the tasks page under the profile menu, you can schedule tasks you want ChatGPT to do, such as send you a weekly global news briefing, get a daily personalized 15-minute workout, or set a reminder for the future.”
“One of my tests of “highly capable AI” is for it to create a visually exciting GLSL shader that runs in Twigl. It involves art, science and coding. No current model, including o1-pro, seems close to pulling that off.” / X
OpenAI quietly funded independent math benchmark before setting record with o3
“There are no rigorous studies yet on how to best prompt reasoning models like o1. They are not chat models, they are designed so that you give them a question and you get an answer. You want to give them context, etc. but we don’t know what the optimal way to do that is yet.” / X
“I’ve been using o1 pro for long context and it makes so many mistakes. It’s not inherently that the code it writes is wrong (it runs/compiles) but it runs incorrectly. The behavior of the code is wrong (breaks certain part of the application). Worst part, it takes so long to” / X
