Benchmark
MathArena Apex
Math benchmarks measure how well models solve maths problems. Scores are percentages. Higher is better.
- Models measured
- 54
- Scores recorded
- 62
- Sources
- 3
- IQ weight
- 2.67
Score distribution
- 26.7%
- Mean
- 19.8%
- Median
- 26.4%
- Std dev
- 0.52%
- 10th percentile
- 65.6%
- 90th percentile
- 0%
- Min
This benchmark has a weight of 2.67 for chat models, measured over 54 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | |
|---|---|---|
| 51 | GLM 4.5· MathArena data | 1% |
| 52 | GPT 5high· MathArena data | 1% |
| 53 | GPT 5.1high· MathArena data | 1% |
| 54 | GPT OSS 120Bhigh· MathArena data | 1% |
| 55 | GPT-5 minihigh· MathArena data | 1% |
| 56 | DeepSeek V3.1think· MathArena data | 0.52% |
| 57 | DeepSeek V3.2 Expthink· MathArena data | 0.52% |
| 58 | Gemini 2.5 Pro· MathArena data | 0.52% |
| 59 | GLM 4.6· MathArena data | 0.52% |
| 60 | Qwen3 30B A3Bthink· MathArena data | 0.52% |
| 61 | Qwen3.5 9B· MathArena data | 0.52% |
| 62 | Kimi K2 Thinking· MathArena data | 0% |
Showing 51–62 of 62 results.
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) · Data by LLM Stats · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.