Mathematical Problem Solving on IMO 2025
83.3PointsDeepSeekMath-V2
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| DeepSeekMath-V2Search Strategy=High-Compute Search, Problems Solved=P1, P2, P3, P4, P52025.11 | 83.3 | — | — | — | — | — | — | — | |
| progressive pessimistic verification (prog@4/4)Backbone Model=Gemini 3 Pro Preview, Verification=progressive pessimistic verification (prog@4/4)2025.11 | 22 | 6 | 0 | 7 | 5 | 4 | 0 | 985 | |
| pessimistic verification (pes@12)Backbone Model=Gemini 3 Pro Preview, Verification=pessimistic verification (pes@12)2025.11 | 17 | 2 | 0 | 6 | 5 | 4 | 0 | 1,310 | |
| directBackbone Model=Gemini 3 Pro Preview, Verification=direct2025.11 | 13 | 2 | 0 | 0 | 5 | 6 | 0 | 129 |