Analysis updated 2026-05-18
Study machine checked proofs of all six 2026 IMO problems written in Lean 4.
Rebuild and reverify the proof libraries against the pinned Mathlib version.
Use the verification script to independently confirm each statement and solution pair is consistent.
| axiommath/imo2026 | lean-dojo/torchlean | gaearon/analysis-solutions | |
|---|---|---|---|
| Stars | 92 | 55 | 42 |
| Language | Lean | Lean | Lean |
| Last pushed | — | — | 2025-09-23 |
| Maintenance | — | — | Quiet |
| Setup difficulty | moderate | hard | moderate |
| Complexity | 4/5 | 5/5 | 4/5 |
| Audience | researcher | researcher | researcher |
Figures from each repo's GitHub metadata at analysis time.
Requires the Lean 4 toolchain and downloading the Mathlib cache before building.
This repository is a record of a result rather than a general purpose tool. It holds the formal problem statements and verified proofs for all six problems from the 2026 International Mathematical Olympiad, one of the world's most prestigious math competitions for pre college students. The proofs were produced automatically by AxiomProver, an AI system built by Axiom Math that writes and checks mathematical proofs using Lean 4, a programming language designed for expressing math so precisely that a computer can confirm a proof is correct rather than just plausible. According to the README, AxiomProver solved all six competition problems and achieved a perfect score. Each problem has its own folder containing two files: one with the formal statement of the problem as generated by the AI, and one with the complete, machine checked solution, also generated by the AI. The README lists how long each solution took to produce and how many lines of Lean code each proof required, ranging from under ten minutes for the easier problems to over fourteen hours for the hardest one. The repository includes instructions for building the proofs against a specific version of the Mathlib math library, and a verification script that rechecks each problem and solution pair for consistency, either using a cached result or forcing a fresh recomputation. This project is aimed at people working in formal mathematics, automated theorem proving, or AI for math research, not general software developers. Using it requires familiarity with Lean 4 and an interest in reading or rebuilding formally verified proofs rather than running an application.
Formal Lean 4 statements and AI-generated, machine-checked proofs for all six 2026 International Mathematical Olympiad problems.
Mainly Lean. The stack also includes Lean, Mathlib, Python.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly researcher.
This repo across BitVibe Labs
Verify against the repo before relying on details.