Benchmark
MathArena 12/2025
Math benchmarks measure how well models solve maths problems. Scores are percentages.
All math benchmarksScores from MathArena data.
- Models measured
- 19
- Scores recorded
- 20
- Sources
- 1
- IQ weight
- 1.24
Score distribution
- 43.2%
- Mean
- 41.5%
- Median
- 9.8%
- Std dev
- 32.4%
- 10th percentile
- 57.4%
- 90th percentile
- 27.9%
- Min
This benchmark has a weight of 1.24 for chat models, measured over 19 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 | Gemini 3.1 Pro Previewhigh· MathArena data | 66.2% |
| 2 | GPT 5.4xhigh· MathArena data | 60.3% |
| 3 | Claude Opus 4.6high· MathArena data | 57.4% |
| 4 | GPT 5.2high· MathArena data | 52.2% |
| 5 | Gemini 3 Pro Previewpreview· MathArena data | 51.1% |
| 6 | Grok 4.1 Fastreasoning· MathArena data | 50% |
| 7 | Gemini 3 Flashhigh· MathArena data | 43.4% |
| 8 | Kimi K2.5high· MathArena data | 41.9% |
| 9 | Step 3.5 Flash· MathArena data | 41.9% |
| 10 | DeepSeek V3.2think· MathArena data | 41.5% |
| 11 | Qwen3.5 27B· MathArena data | 41.2% |
| 12 | Qwen3.5 35B A3B· MathArena data | 39.7% |
| 13 | Qwen3.5 9B· MathArena data | 39.7% |
| 14 | GLM 5· MathArena data | 38.2% |
| 15 | Qwen3.5 397B A17B· MathArena data | 38.2% |
| 16 | Qwen3.5 4B· MathArena data | 34.2% |
| 17 | Nemotron 3 Super· MathArena data | 33.8% |
| 18 | GPT 5.2low· MathArena data | 32.4% |
| 19 | Qwen3 4Bthink· MathArena data | 32.4% |
| 20 | Qwen3 30B A3Bthink· MathArena data | 27.9% |
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)