Net Good IndexSubmit a correction

Capability Ledger · verified · Mathematics

Gemini Deep Think reaches IMO 2025 gold-medal standard

In July 2025, Google DeepMind’s Gemini Deep Think was officially graded at 35/42 on the IMO 2025 paper under contest-style conditions.

Tier 3 MajorMethodology 0.2
Submit a correction

Capability score

1.58

Does not count toward Net Good or the leaderboard.

10 base · Major (tier 3 of 5, 10 pts)
× 0.9000 attribution · Primary causal contribution
× 0.5000 evidence · External expert evaluation
× 0.5000 realization · Complete public manuscript
× 0.7000 durability · Useful for many years
Event-level product before credit split: 1.58

Officially graded gold-threshold IMO performance is a major capability milestone (tier 3), not Net Good. Independent OpenAI performance is scored separately rather than 50/50. High attribution. Evidence is coordinator grading plus journalism (0.50).

What happened

IMO coordinators graded an advanced Gemini Deep Think system at 35 of 42 on the 2025 International Mathematical Olympiad paper: five of six problems under contest-style 4.5-hour sessions and natural-language proofs. The result is a capability demonstration, not an official contest medal and not a contribution of new theorems. OpenAI independently reported the same score from a separate experimental model; that is a different event. Human gold medalists still exist above and at this score. This does not count toward Net Good.

Model attribution

1.58

Gemini

Officially graded IMO 2025 solutions scoring 35/42.

Independent system with IMO coordinator grading. OpenAI’s parallel result is a separate event.

Attribution 0.9000 · Credit share 100% · Google DeepMind

Claims

  • Gemini Deep Think scored 35/42 on IMO 2025 as graded by IMO coordinators.

    outcome · supported

  • Gemini officially won an IMO gold medal as a contestant.

    significance · disputed

Sources

primary sources

independent sources

Secondary domains: Computer Science

Revision history

  • 13 Sep 2026 · 0.00 → 1.53

    Initial adjudicated seed score under methodology 0.1.

  • 14 Sep 2026 · 1.53 → 1.58

    Methodology 0.2: Gemini IMO 2025 result is an independent capability event with full credit; OpenAI is scored separately.

  • 17 Sep 2026 · 1.58 → 1.58

    Methodology 0.2: Gemini IMO 2025 result is an independent capability event with full credit; OpenAI is scored separately.

  • 18 Sep 2026 · 1.58 → 1.58

    Methodology 0.2: Gemini IMO 2025 result is an independent capability event with full credit; OpenAI is scored separately.

  • 19 Sep 2026 · 1.58 → 1.58

    Methodology 0.2: Gemini IMO 2025 result is an independent capability event with full credit; OpenAI is scored separately.

  • 21 Sep 2026 · 1.58 → 1.58

    Methodology 0.2: Gemini IMO 2025 result is an independent capability event with full credit; OpenAI is scored separately.

  • 6 Oct 2026 · 1.58 → 1.58

    Methodology 0.2: Gemini IMO 2025 result is an independent capability event with full credit; OpenAI is scored separately.

  • 7 Oct 2026 · 1.58 → 1.58

    Methodology 0.2: Gemini IMO 2025 result is an independent capability event with full credit; OpenAI is scored separately.

  • 11 Oct 2026 · 1.58 → 1.58

    Methodology 0.2: Gemini IMO 2025 result is an independent capability event with full credit; OpenAI is scored separately.

  • 11 Oct 2026 · 1.58 → 1.58

    Methodology 0.2: Gemini IMO 2025 result is an independent capability event with full credit; OpenAI is scored separately.

  • 21 Oct 2026 · 1.58 → 1.58

    Methodology 0.2: Gemini IMO 2025 result is an independent capability event with full credit; OpenAI is scored separately.

  • 24 Oct 2026 · 1.58 → 1.58

    Methodology 0.2: Gemini IMO 2025 result is an independent capability event with full credit; OpenAI is scored separately.

  • 27 Oct 2026 · 1.58 → 1.58

    Methodology 0.2: Gemini IMO 2025 result is an independent capability event with full credit; OpenAI is scored separately.

  • 28 Oct 2026 · 1.58 → 1.58

    Methodology 0.2: Gemini IMO 2025 result is an independent capability event with full credit; OpenAI is scored separately.

Submit a correction