The Starting Point for Reasoning
Joined September 2025
- Tweets1.6K
- Following71
- Followers24.3K
- Likes1.1K
Pinned Tweet
Most think vibecoding at scale is slop. Here’s a counterexample.
Today we release a 16M-line formalization of the entire Stacks Project, critical infra underlying modern math, with optimized code quality and its highly interconnected structure preserved.
A mathematical city 🌆
Axiom retweeted
Axiom Math announced it has formally verified the Stacks Project, one of the largest bodies of mathematics ever written by humans. A fleet of AI agents translated the library’s roughly 7,654 pages into about 16 million lines of computer-checked code. techcrunch.com/2026/11/05/ax…
Axiom retweeted
Congrats on Stacks, and on closing the 4.35.0 toolchain gap in two days!
Genuine question: has anyone pointed this at non-math software, like a lease/lock protocol with a mutual-exclusion goal?
Curious whether the core/frontier split survives contact with concurrency.
Axiom retweeted
🎉 the solution to the moving sofa problem, given by Jineon Baek, is autoformalized in Lean! 66,000 lines of code written in 4 days, passes comparator
repo here: github.com/deancureton/Movin…
🧵
Axiom retweeted
Axiom Math wants to keep AI slop out of mathematics. ~ Aditya Mehta.
techcrunch.com/2026/11/05/ax…
#LeanProver #ITP #AI4Math
Axiom retweeted
i’m bullish on formalization. so glad @axiomai is doing this. we owe it to math.
Today Axiom formalized the proofs of the celebrated Weil conjectures, foundational pillars of algebraic geometry.
This took us less than a day.
Had the Stacks project not been formalized with high code quality, this success would've not been possible.
What we hope to show is that Stacks is a reusable, foundational library.
With the underlying theory built the right way, formalization of acclaimed results naturally fell, just like "a land mass being silently swallowed by the rising sea."
This is just the first example of such paradigm shift. From machine-assisted *formal proving* of individual theorems, to autonomous swarms **library building** of entire theories.
We are witnessing Grothendieck's vision come to life in Lean.
a first-order logic thinker 🤝 a first-order logic programming language
Axiom retweeted
Tune in to hear this Giga Brain talk
Excited to be speaking next week at SGLang Summit by @lmsysorg on how we've built scalable post-training infrastructure to push the frontier of math
Axiom retweeted
Excited to be speaking next week at SGLang Summit by @lmsysorg on how we've built scalable post-training infrastructure to push the frontier of math
An in-house training stack for mathematical formalization.
Tobias Gessler from @axiommathai shares how the team approaches building training infrastructure for this specialized domain.
Axiom retweeted
Axiom just stress-tested its Stacks Lean library the only way that matters: by building the next proof on it.
Weil conjectures -> Lean
All four classical statements - including the finite-field Riemann hypothesis - formalized in under a day once the foundations were in place.
I'd rather watch a library earn its keep than collect another "AI did math" headline.
Via @axiommathai #AI4Science #FormalMath
Axiom retweeted
It just keeps getting more insane. These are foundational pillars of algebraic geometry… And they say I took them one day.
Axiom retweeted
Thanks a lot for all of your work, of the entire team - and I also thank all those who contributed on Stacks - Your work is saving us so much trouble... Thank you!
Axiom retweeted
We need something similar for biology!
Axiom retweeted
Formal-maxxing
Today Axiom formalized the proofs of the celebrated Weil conjectures, foundational pillars of algebraic geometry.
This took us less than a day.
Had the Stacks project not been formalized with high code quality, this success would've not been possible.
What we hope to show is that Stacks is a reusable, foundational library.
With the underlying theory built the right way, formalization of acclaimed results naturally fell, just like "a land mass being silently swallowed by the rising sea."
This is just the first example of such paradigm shift. From machine-assisted *formal proving* of individual theorems, to autonomous swarms **library building** of entire theories.
We are witnessing Grothendieck's vision come to life in Lean.
a first-order logic thinker 🤝 a first-order logic programming language
Post-train a formal theorem proving model to reach frontier performance to push the frontier of math 🔥🔥🔥
An in-house training stack for mathematical formalization.
Tobias Gessler from @axiommathai shares how the team approaches building training infrastructure for this specialized domain.
Axiom retweeted
An in-house training stack for mathematical formalization.
Tobias Gessler from @axiommathai shares how the team approaches building training infrastructure for this specialized domain.
Axiom retweeted
Fields medalist Vladimir Voevodsky explained why formalization matters:
Mathematics has become too complex to rely solely on human verification. He cited errors in his own works that were undetected for decades.
ias.edu/ideas/2014/voevodsky…
