IMO 2025
DeepSeekMath-V2 83.3% leads since Nov 2025 Fading Models write full proofs for the six problems of the 2025 International Mathematical Olympiad. Graders score proofs out of 42 points, mapped to medal cutoffs.
This is a one-time competition exam, not a maintained benchmark.
- Level
- Competition The strongest students, in olympiad-style contests
- Scored
- Graded by people Human experts grade or compare the outputs.
- Answers
- Written proof
4 models share the top score of 83.3% among models released by Jun 2026
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
The newest model with a score here came out in Jun 2026, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6 Astra
- Claude Opus 5.5
- GPT-6.1 Terra
- GPT-6.1 Sol
- and 16 more
- 1 DeepSeekMath-V2 83.3%
- 2
Seed2.0 Pro 83.3% - 3
Nemotron-Cascade 2 30B-A3B 83.3% - 4 Nemotron 3 Ultra 83.3%
- 5 Seed2.1 Pro 65.2%
- 6
Gemini 2.5 Deep Think 60.7%
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | DeepSeekMath-V2 | 83.3% | Nov 2025 | Model card | |
| 2 | Seed2.0 Pro | 83.3% | Feb 2026 | Model card | |
| 3 | Nemotron-Cascade 2 30B-A3B | 83.3% | Mar 2026 | Model card | |
| 4 | Nemotron 3 Ultra | 83.3% | Jun 2026 | Model card | |
| 5 | Seed2.1 Pro | 65.2% | Jun 2026 | Model card | |
| 6 | Gemini 2.5 Deep Think | 60.7% | Aug 2025 | Model card |
The benchmark's own leaderboard has 7 entries, 6 of them models we track. We plot every source in the charts below.
2 models have held number one since its release in Jul 2025
Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for the benchmark’s own leaderboard and what labs claim in their cards.
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Labs headline it less often now, last in Jun 2026
Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.
| Date | Launch post | Score | As the post writes it |
|---|---|---|---|
| 1 Jun 2026 |  MiniMax-M3 | 35 | IMO 2025 |
| 19 Mar 2026 | Nemotron-Cascade 2 30B-A3B | 35 | IMO 2025 (35 pts, gold medal level) |
| 14 Feb 2026 | Seed2.0 Pro | 35 | IMO 2025, 35/42, gold (Table 2) |
| 12 Feb 2026 | Gemini 3 Deep Think | International Math Olympiad 2025, gold-medal level | |
| 10 Feb 2026 |  Ring-2.5-1T | 35 | IMO 2025 (full score 42), self-tested |
| 19 Dec 2025 | Seed-Prover 1.5 | 35 | IMO 2025, Lean proofs for P1 to P5 within 16.5 hours |
| 27 Nov 2025 | DeepSeekMath-V2 | 83.3 | IMO 2025 Points (P1 to P5 solved) |
| 1 Aug 2025 | Gemini 2.5 Deep Think | 60.7 | IMO 2025 (Bronze medal grade; pass@1) |
| 31 Jul 2025 | Seed-Prover | 4 | IMO 2025 4/6 (Heavy, 5/6 post-competition) |
7 system cards report it, from 4 labs
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 23 Jun 2026 | Seed2.1 Pro model card | 65.2 | IMO 2025, Seed2.1 Pro column (Table 7). Deep Think 81.0 | Capabilities |
| 9 Jun 2026 | Nemotron 3 Ultra technical report | 83.3 | IMO 2025, 35/42, generate-verify-refine test-time scaling, human graded (Table 11) | Capabilities |
| 19 Mar 2026 | Nemotron-Cascade 2 30B-A3B technical report | 83.3 | IMO 2025, 35/42 points, gold, human expert graded except P2 (Tables 1 and 2), recorded as 83.3% | Capabilities |
| 14 Feb 2026 | Seed2.0 Pro model card | 83.3 | IMO 2025, 35/42 points, gold, recorded as 83.3% (Table 2) | Capabilities |
| 1 Dec 2025 | DeepSeek-V3.2-Speciale technical report | 35 | IMO 2025, 35/42 points, gold medal (Table 4) | Capabilities |
| 27 Nov 2025 | DeepSeekMath-V2 technical report | 83.3 | IMO 2025 Points, P1 to P5 solved, high-compute search (Table 1) | Capabilities |
| 21 Oct 2025 | Ring-1T technical report | IMO 2025 in the AWorld multi-agent framework, natural language only, solved Problems 1, 3, 4 and 5 on the first attempt, silver medal level, no point total (Section 3.3) | Capabilities |
Versions
More in Math
- IMO-AnswerBench In use 400 olympiad problems with short verifiable answers, robustified against memorization. Scored by answer match. GLM-5.2 91%
- AIME 2026 In use Models solve the 30 integer-answer problems from the 2026 AIME I and II math competitions. Scored by exact answer match. GPT-5.5 100%
- HMMT 2026 In use Models solve the 33 final-answer problems from the February 2026 Harvard-MIT Mathematics Tournament. Scored by exact answer match. GPT-5.5 98.5%
- MathArena (Apex, IMO, USAMO) In use Evaluates models on newly released math competitions after release. Final-answer or human/LLM-judged proof scores. Claude Opus 4.8 81.3%
- PolyMath In use Models solve 9,000 parallel math problems in 18 languages across four difficulty levels, from school math to frontier math. Final answers are checked.
Qwen3.7-Max 86.5% - USAMO 2026 In use Models write full proofs for the six problems of the 2026 USA Mathematical Olympiad. Judges grade each proof from 0 to 7 against rubrics. GPT-5.5 98.2%