AI Judge changed title after evaluation, original title: "Axiom Math says its AxiomProver generated formal proofs for eight arXiv papers, with five accepted by peer-reviewed journals"
One accepted paper credits AxiomProver for automating its proofs
Exclusive: Math AI startup can prove its work https://www.axios.com/2026/05/26/axiom-ai-math-journal?utm_campaign=mrf-utm_campaign=editorial&utm_source=x&utm_medium=owned_social&utm_source=twitter&utm_medium=social&mrfcid=202605266a0a8ec19dec6c1d889c0d6d
Users are celebrating AxiomProver generating formal proofs for math papers accepted in journals because verifiable AI output builds trust and enables reliable advances in research and formalization.
No Digg Deeper questions have been answered for this story yet.
1/ Since February, 8 papers across algebraic geometry, representation theory, number theory, combinatorics have been quietly appearing on arXiv.
Proofs by AxiomProver.
5 papers are now accepted at solid peer-reviewed math journals. To our knowledge, a first for the literature.
Incredible progress in Math+AI. Honestly what a great time to be a mathematician when much of the technical burden is lifted by machines of loving grace.
1/ Since February, 8 papers across algebraic geometry, representation theory, number theory, combinatorics have been quietly appearing on arXiv.
Proofs by AxiomProver.
5 papers are now accepted at solid peer-reviewed math journals. To our knowledge, a first for the literature.
AxiomProver quietly publishes a handful of papers in peer-reviewed math journals
1/ Since February, 8 papers across algebraic geometry, representation theory, number theory, combinatorics have been quietly appearing on arXiv.
Proofs by AxiomProver.
5 papers are now accepted at solid peer-reviewed math journals. To our knowledge, a first for the literature.
We keep building. We keep our heads down. Rarely we pause and reflect. This is a moment we do:
In 100 days, we have 8 papers live, 6 in the pipeline - on average one paper in 1-2 weeks.
These papers solve open problems posed by renowned mathematicians, output verified in Lean.
2/ We sat down with @inafried of @axios on AxiomProver's first 100 days as an AI mathematician -- and discussed how our AI went from first-time competing at math Olympiads (Putnam) to publishing journal papers in under four months.
https://www.axios.com/2026/05/26/axiom-ai-math-journal
One more journal publication live just now
1/ Since February, 8 papers across algebraic geometry, representation theory, number theory, combinatorics have been quietly appearing on arXiv.
Proofs by AxiomProver.
5 papers are now accepted at solid peer-reviewed math journals. To our knowledge, a first for the literature.
We talk a lot about developer productivity and experience (DX) for software engineering, but researcher productivity (RX) is also having a major moment.
Combining automated drivers with collaborative open-source libraries like Mathlib is unlocking so much latent potential in mathematics research, love to see it:
1/ Since February, 8 papers across algebraic geometry, representation theory, number theory, combinatorics have been quietly appearing on arXiv.
Proofs by AxiomProver.
5 papers are now accepted at solid peer-reviewed math journals. To our knowledge, a first for the literature.
Congrats to our math team. We've tried unlocking the room we keep them in, but at this point they're having so much fun they won't come out.
1/ Since February, 8 papers across algebraic geometry, representation theory, number theory, combinatorics have been quietly appearing on arXiv.
Proofs by AxiomProver.
5 papers are now accepted at solid peer-reviewed math journals. To our knowledge, a first for the literature.
2/ We sat down with @inafried of @axios on AxiomProver's first 100 days as an AI mathematician -- and discussed how our AI went from first-time competing at math Olympiads (Putnam) to publishing journal papers in under four months.
https://www.axios.com/2026/05/26/axiom-ai-math-journal
3/ Math journal review cycles can famously take years. Why so fast now?
Every proof is generated in machine-verified Lean, then paired with a human exposition. The mathematician authors are there to explain the theorem, not to prove it.
AxiomProver produces math one can trust.
4/ Solve cycles are fast too: a mathematician sends an open research problem at 10am and gets a proof back at 4pm the same day.
Examples are some deep, elegant results in number theory: 100% of primes are partially regular, and (under abc) Ramanujan's tau misses 100% of primes.
It is happening....
1/ Since February, 8 papers across algebraic geometry, representation theory, number theory, combinatorics have been quietly appearing on arXiv.
Proofs by AxiomProver.
5 papers are now accepted at solid peer-reviewed math journals. To our knowledge, a first for the literature.
Great to see these maths papers getting published and accepted by the community. Automated proofs are becoming mainstream!
1/ Since February, 8 papers across algebraic geometry, representation theory, number theory, combinatorics have been quietly appearing on arXiv.
Proofs by AxiomProver.
5 papers are now accepted at solid peer-reviewed math journals. To our knowledge, a first for the literature.
5/ Already exciting to see AxiomProver in Advances in Applied Mathematics, Archiv der Mathematik, Indagationes Mathematicae, and more:
https://axiommath.ai/papers
From formal theorem proving to autoformalization to conjecturing, math and science will be accelerated by Verified AI!
Interesting idea on prime "lanes" mod 30—yes, every prime >5 falls into exactly those 8 coprime residues, per basic sieving (Dirichlet guarantees inf many in each). Your Notebook 19 transition heatmap and 23's graph/Laplacian embeddings on lane trajectories look like a solid way to visualize persistence and similarities in distributions.
AxiomProver's verified Lean proofs + human exposition is impressive for trustable AI math. Could this architecture help generate testable conjectures there? Curious to see more. 📐
@CarinaLHong How do I get access to AxiomProver? Sick of chatting back and forth with ChatGPT 5.5 Pro and getting nowhere on a problem. Help!!!!
This is so exciting!
AxiomProver quietly publishes a handful of papers in peer-reviewed math journals
@axiommathai @grok in english, was ramanujan wrong?
@axiommathai 🔥 🔥 🔥
Congrats on this accomplishment!
@axiommathai the pipeline detail that matters: 'produced automatically in Lean/Mathlib by AxiomProver from a natural-language statement.' formalization has always been the expensive step, not proof discovery. if that gap is automated, the cost of publishable verified math drops dramatically
@DimitrisPapail if AI is lifting the technical burden in math, the next bottleneck will be peer review keeping up with the volume of provable results
