Benchmark
MathArena Kangaroo 2025 1-2
Math benchmarks measure how well models solve maths problems. Scores are percentages.
All math benchmarksScores from MathArena data.
- Models measured
- 21
- Scores recorded
- 21
- Sources
- 1
- IQ weight
- 0.99
Score distribution
- 75.4%
- Mean
- 76%
- Median
- 12.3%
- Std dev
- 61.5%
- 10th percentile
- 93.8%
- 90th percentile
- 59.4%
- Min
This benchmark has a weight of 0.99 for chat models, measured over 21 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 | 95.8% |
| 2 | GPT 5.4xhigh· MathArena data | 94.8% |
| 3 | Kimi K3max· MathArena data | 93.8% |
| 4 | Gemini 3.5 Flashhigh· MathArena data | 89.6% |
| 5 | Gemini 3 Flashhigh· MathArena data | 87.5% |
| 6 | Gemini 3.1 Pro Previewhigh· MathArena data | 86.5% |
| 7 | Gemini 3.6 Flash· MathArena data | 83.3% |
| 8 | Qwen3.5 397B A17B· MathArena data | 81.3% |
| 9 | GPT 5.2high· MathArena data | 80.2% |
| 10 | Gemini 3 Pro Previewpreview· MathArena data | 76% |
| 11 | Kimi K2.5high· MathArena data | 76% |
| 12 | Claude Opus 4.8max· MathArena data | 70.8% |
| 13 | GPT 5high· MathArena data | 68.8% |
| 14 | GLM 4.5V· MathArena data | 65.6% |
| 15 | GPT 5.1high· MathArena data | 65.6% |
| 16 | Gemini 2.5 Pro· MathArena data | 64.6% |
| 17 | Claude Sonnet 4.5think· MathArena data | 61.5% |
| 18 | GPT-5 minihigh· MathArena data | 61.5% |
| 19 | Grok 4· MathArena data | 61.5% |
| 20 | Grok 4.1 Fastreasoning· MathArena data | 60.4% |
| 21 | Claude Opus 4.6high· MathArena data | 59.4% |
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)