An advanced benchmark for evaluating complete and logically rigorous proofs generated for difficult Olympiad-level mathematics problems.
Browse the latest scores, model modes, release dates, and parameter sizes for IMO-ProofBench Advanced.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
| Rank | Model | License | |||
|---|---|---|---|---|---|
 DeepSeekMath-V2 Thinking Level · High | 61.90 | 2025-11-27 | 685B | Free Commercial | |
 GPT-5 Thinking Enabled | 20.00 | 2025-08-07 | Unknown | Closed | |
 Grok 4 Thinking Enabled | 18.60 | 2025-07-10 | Unknown | Closed | |
4 |  Gemini 2.5-Pro Thinking Enabled | 17.60 | 2025-06-05 | Unknown | Closed |
5 |  Qwen3-235B-A22B-Thinking Thinking Enabled | 5.20 | 2025-07-30 | 30.5B | Free Commercial |
6 |  Claude Sonnet 4 Thinking Enabled | 4.80 | 2025-05-23 | Unknown | Closed |
7 | Claude Sonnet 4.5 Thinking Enabled | 4.80 | 2025-09-30 | Unknown | Closed |
8 | DeepSeek-R1-0528 Thinking Enabled | 3.80 | 2025-05-28 | 671B | Free Commercial |
