Benchmark
MathArena USAMO 2026
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
- 0.99
Score distribution
- 60.3%
- Mean
- 51.2%
- Median
- 22.5%
- Std dev
- 36.3%
- 10th percentile
- 95.2%
- 90th percentile
- 35.1%
- Min
This benchmark has a weight of 0.99 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 | GPT 5.5xhigh· MathArena data | 98.2% |
| 2 | GPT 5.4xhigh· MathArena data | 95.2% |
| 3 | Gemini 3.1 Pro Previewhigh· MathArena data | 74.4% |
| 4 | DeepSeek V4 Promax· MathArena data | 60.7% |
| 5 | Kimi K2.6high· MathArena data | 51.2% |
| 6 | Claude Opus 4.6high· MathArena data | 47% |
| 7 | Step 3.5 Flash· MathArena data | 44.6% |
| 8 | Qwen3.5 397B A17B· MathArena data | 36.3% |
| 9 | GLM 5· MathArena data | 35.1% |
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)