Congratulations Levent, Tristan, and OpenAI! What a miraculous time to be alive! I view this as the first successful achievement of Recursive Self Improvement or RSI. Models get increasingly better at Math and get increasingly used by mathematicians to solve all kinds of difficult problems in their domain, thereby contributing very rare and very hard tokens that are then used in the model training to further improve the model's math capabilities: generalizing and drawing connections across sessions and everything else the models learn from; which is then served back, and so on and so forth. This happened first for math but will eventually happen for every domain.

Surprised to learn today how many people did not know that if you don't turn off "Improve the model for everyone" in chat then your data will very likely be trained on. Also, its on by default. Same applies to codex.

Fun fact: RLSlow was named after Thinking, Fast and Slow. The idea was that language models already had a kind of "fast" thinking, producing an answer immediately, and that we could use RL to teach them "slow" thinking: deliberate, multi-token reasoning that spends more compute working through a problem. What I remember most from those days is how early the team developed real conviction in the direction, and how much work went into earning it. We were developing the algorithms, designing careful experiments to test the ideas, and watching the empirical evidence accumulate. These were many long nights babysitting runs, understanding what the results were telling us, and figuring out what to try next. A lot of those early discussions were with Ilya, and later with Jakub. Pretty early, the evidence had already pushed us to a strong view: RL for reasoning would scale. Models would learn to spend more compute at inference time to reason through increasingly hard problems, and this would fundamentally change how we think about inference. Many of us also spent countless hours reading through reasoning traces. Seeing how models arrived at answers, not just the answers themselves, felt like a powerful new lens on generalization and alignment. These ideas feel obvious now. They really weren’t then.

Let me take this opportunity to pay tribute to the RLSlow team :) It was a priviledge to lead it through the years, first with @ilyasut then with @merettm and finally on my own - the best team you can ever wish for. It feels like ages ago when we started working on the foundations of reasoning (later berry) with @TrapitBansal, Francis Song, @_smileyball and @sidorszymon. Many more things happened since then, huge thank you to all teammates through the years: Peter Zhokhov, @unsafePerformIO @vladi_petroff @masonmeyer_ @afrozenator @lukasz_kondr @dmdohan @wang12tao @TheRealRPuri @edwardjhu @eqhylxx @suchirbalaji and Dan Selsam for all the help. I hope this great technical work continues to be truly and broadly beneficial.

Excited to see Muse Spark 1.3 out in the world. There’s a lot of progress in this release, and with 1.3 Max specifically, the intelligence you get per dollar is kind of ridiculous! If you’re a developer and haven’t tried Muse Spark yet, you’re running out of reasons not to.

Since @shuchaobi is posting pictures here’s one. Fun fact — we literally went to the beach when the model was solving IChO. We trusted it enough. This pic was taken then. Over the years it’s stuff like this that one remembers :) I wonder if other labs got their Golds with the same panache 😇 (with @TrapitBansal)

Students who earn gold at these Olympiads are exceptional and often go on to make major contributions in their fields. As we continue to scale these models, it’s exciting to imagine what one that excels across physics, chemistry, and mathematics could achieve. 3/5

We have been investing in advancing multimodal reasoning for Muse Spark and we participated in this year's major science olympiads. It was great to see our model obtain a 100th percentile score at the International Physics Olympiad (🥇), International Chemistry Olympiad (🥇) and the Asian Physics Olympiad (🥇) in theory exams!

Students who earn gold at these Olympiads are exceptional and often go on to make major contributions in their fields. As we continue to scale these models, it’s exciting to imagine what one that excels across physics, chemistry, and mathematics could achieve. 3/5

Every parent I know here has the same problem: finding something to do with their kids and dog that still feels new, while threading limited weekend hours, drive time, weather, and nap schedules. It's basically a second job! I wanted my sacred family time spent making memories, not scrolling for plans. So I built : one plain sentence in, a whole day out. kids, dogs, naps, all of it. It's live, go plan something 🎠

Introducing Phirky: your bay area weekend planner. Type it like you'd say it: "rainy Saturday, toddler under 3, under $20, dog-friendly, near Palo Alto"; and hands you a ranked plan. Every place human-verified: realistic hours, parking, shade, water bowls.