The Bitter lesson does not say to not bother with methods research. It says to not bother with methods that are handcrafted datapoints in disguise.

story wouldn't be complete without a shoutout to hunter Karl Cobbe Vineet Kosaraju for writing the second and more usable (in more ways than one) version!

in almost all cases I agree with those who say the distinction between researcher and engineer is meaningless, but it's hard to think of jakub as anything but an exceptional researcher, despite him being as good an engineer as any in the company

people will literally build superintelligence to do their taxes instead of properly funding the irs

☕️🥯

gpt-oss is our new open-weight model family! the bigger one runs on a single GPU, you can run the small one on your laptop. Go install it right now, seriously! Telling your laptop to do something and watching it happen made me feel the AGI like nothing since ChatGPT.

mistakes were made

there was a time when aime looked like a pretty tall mountain to climb

many plots today made me sad. this plot is important though: the models are truly getting more and more practical in the real-world

gpt-5 is here today! so excited to see what you'll do with it. we've focused a ton on coding in particular, and i think you'll really enjoy using it for the hardest problems and for building web apps end to end. what you can get out of a single prompt is really remarkable

fun to share more about agents and talk about research at openai with christina kim! thank you Erik Torenberg and Sarah Wang for having us on the pod

1/ I competed for Team USA at IOI in 2015, so this achievement hits home for me. The biggest highlight: we *did not* train a model specifically for IOI. Our IMO gold model actually set a new state of the art in our internal competitive programming evals. Reasoning generalizes!

My first project at OpenAI involved teaching our models to reason and use tools by improving their competitive programming skills. Back then, GPT-4 struggled with even the simplest Codeforces problems, often oom-ing in the sandbox. It's incredible to see that just 2.5 years

just say no to handcrafted tt strategies (unless you have to) (but really, no)

Announcing gold medal-level performance (#6 relative to humans; #1 relative to AIs) on this year’s International Olympiad on Informatics (IOI). Achieved with an AI reasoning system, without IOI specific training!

Just two years ago, our smartest models could barely solve the easiest competitive programming problems. Last week, our latest reasoning models achieved a gold medal score at the International Olympiads of Informatics. Competitive programming is one of the cleanest examples of

Just 2 years ago this line of work could barely pass easy CF tasks. Today our models reach IOI gold level on the official AI track, first among AI entrants. Long project with Ahmed El-Kishky Lorenz Kuhn Alexander Wei Borys Minaiev Oleg Mürk Mostafa Rohaninejad Ignasi Clavera. See Sheryl Hsu thread.

you don't want the truth because deep down in places you don't talk about at parties, you want me on that wall -- you need me on that wall