Net Good IndexSubmit a correction

Capability Ledger · verified · Mathematics

AlphaGeometry solves olympiad plane-geometry problems without human proof data

DeepMind’s neuro-symbolic AlphaGeometry solved 25 of 30 IMO-AG-30 geometry problems in a Nature paper, approaching average human gold-medalist performance on that set.

Tier 3 MajorMethodology 0.2
Submit a correction

Capability score

2.77

Does not count toward Net Good or the leaderboard.

10 base · Major (tier 3 of 5, 10 pts)
× 0.9000 attribution · Primary causal contribution
× 0.7000 evidence · Peer review or substantial independent validation
× 0.5500 realization · Complete public manuscript
× 0.8000 durability · Useful for many years
Event-level product before credit split: 2.77

Peer-reviewed olympiad-geometry performance is a major automated-reasoning capability (tier 3). It stays on the Capability Ledger because the problems were already solved mathematics. High attribution. Evidence is Nature. Realization is demonstrated on a historical set.

What happened

On 17 January 2024, Nature published AlphaGeometry, a language model trained on synthetic geometry theorems that guides a symbolic deduction engine. On 30 recent olympiad-level plane-geometry problems it solved 25, versus 10 for a prior automated baseline. Human-readable proofs were produced; former olympiad medalist Evan Chen reviewed examples. The system covers only Euclidean plane geometry, not a full IMO paper. This is a capability milestone on a historical contest set whose answers already existed. It is not newly created mathematical knowledge and does not count toward Net Good.

Model attribution

2.77

AlphaGeometry

Neuro-symbolic prover that proposed auxiliary constructions for a symbolic engine.

The synthetic-data language model plus deduction engine is the published system.

Attribution 0.9000 · Credit share 100% · Google DeepMind

Claims

  • AlphaGeometry solved 25 of 30 problems in the IMO-AG-30 geometry set.

    outcome · supported

  • AlphaGeometry by itself earned an IMO gold medal in a live contest.

    outcome · disputed

Sources

primary sources

Secondary domains: Computer Science

Revision history

  • 13 Sep 2026 · 0.00 → 2.77

    Initial adjudicated seed score under methodology 0.1.

  • 14 Sep 2026 · 2.77 → 2.77

    Methodology 0.2: olympiad benchmark moved to the Capability Ledger; not Net Good.

  • 17 Sep 2026 · 2.77 → 2.77

    Methodology 0.2: olympiad benchmark moved to the Capability Ledger; not Net Good.

  • 18 Sep 2026 · 2.77 → 2.77

    Methodology 0.2: olympiad benchmark moved to the Capability Ledger; not Net Good.

  • 19 Sep 2026 · 2.77 → 2.77

    Methodology 0.2: olympiad benchmark moved to the Capability Ledger; not Net Good.

  • 21 Sep 2026 · 2.77 → 2.77

    Methodology 0.2: olympiad benchmark moved to the Capability Ledger; not Net Good.

  • 6 Oct 2026 · 2.77 → 2.77

    Methodology 0.2: olympiad benchmark moved to the Capability Ledger; not Net Good.

  • 7 Oct 2026 · 2.77 → 2.77

    Methodology 0.2: olympiad benchmark moved to the Capability Ledger; not Net Good.

  • 11 Oct 2026 · 2.77 → 2.77

    Methodology 0.2: olympiad benchmark moved to the Capability Ledger; not Net Good.

  • 11 Oct 2026 · 2.77 → 2.77

    Methodology 0.2: olympiad benchmark moved to the Capability Ledger; not Net Good.

  • 21 Oct 2026 · 2.77 → 2.77

    Methodology 0.2: olympiad benchmark moved to the Capability Ledger; not Net Good.

  • 24 Oct 2026 · 2.77 → 2.77

    Methodology 0.2: olympiad benchmark moved to the Capability Ledger; not Net Good.

  • 27 Oct 2026 · 2.77 → 2.77

    Methodology 0.2: olympiad benchmark moved to the Capability Ledger; not Net Good.

  • 28 Oct 2026 · 2.77 → 2.77

    Methodology 0.2: olympiad benchmark moved to the Capability Ledger; not Net Good.

Submit a correction