Benchmark
MathArena USAMO 2025
Math benchmarks measure how well models solve maths problems. Scores are percentages.
All math benchmarksScores from MathArena data.
- Models measured
- 9
- Scores recorded
- 9
- Sources
- 1
- IQ weight
- 1.22
Score distribution
- 10.4%
- Mean
- 4.2%
- Median
- 10.3%
- Std dev
- 2.8%
- 10th percentile
- 24.4%
- 90th percentile
- 2.1%
- Min
This benchmark has a weight of 1.22 for chat models, measured over 9 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | |
|---|---|---|
| 1 | DeepSeek R1· MathArena data | 30.1% |
| 2 | Gemini 2.5 Pro· MathArena data | 24.4% |
| 3 | o4 Minihigh· MathArena data | 19.1% |
| 4 | Grok 3think· MathArena data | 4.8% |
| 5 | Gemini 2.0 Flashthinking· MathArena data | 4.2% |
| 6 | Claude Sonnet 3.7think· MathArena data | 3.7% |
| 7 | QwQ 32B· MathArena data | 3% |
| 8 | O1 Prohigh· MathArena data | 2.8% |
| 9 | o3 Minihigh· MathArena data | 2.1% |
Tokens the model generated per task, as measured by the source. Comparable within this benchmark only.
Data credits: MathArena (ETH SRI / INSAIT), MIT — cite dekoninck2026matharena (licence)