math benchmark
MathArena Apex
MathArena Apex is a challenging math contest benchmark featuring the most difficult mathematical problems designed to test advanced reasoning and problem-solving abilities of AI models. It focuses on olympiad-level mathematics and complex multi-step mathematical reasoning.
Updated Aug 11, 2026
Models7
Model coverage7
MetricScore
EvidenceB
MathArena Apex Ranking
Higher score ranks better on this benchmark.
| 1 | DE | 90.2% | 100.0% | 7 | C | |
| 2 | DE | 85.7% | 83.3% | 7 | C | |
| 3 | AC | 44.5% | 66.7% | 7 | C | |
| 4 | TE | 38.7% | 50.0% | 7 | C | |
| 5 | BY | 35.4% | 33.3% | 7 | C | |
| 6 | BY | 31.3% | 16.7% | 7 | C | |
| 7 | GO | 23.4% | 0.0% | 7 | C |
MathArena Apex Score Distribution
A closer view of the leading scores on this benchmark.
MathArena Apex Highlights
The leading models and scores on this benchmark.
What is MathArena Apex?
What MathArena Apex measures and how its scores work.
MathArena Apex is a challenging math contest benchmark featuring the most difficult mathematical problems designed to test advanced reasoning and problem-solving abilities of AI models. It focuses on olympiad-level mathematics and complex multi-step mathematical reasoning.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- MathArena Apex
- Modality
- text
- Primary category
- math
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- matharena-apex|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about MathArena Apex.
Which model scores highest on MathArena Apex?
DeepSeek-V4-Pro-Max is currently ranked first with 90.2%.
What does MathArena Apex measure?
MathArena Apex is a challenging math contest benchmark featuring the most difficult mathematical problems designed to test advanced reasoning and problem-solving abilities of AI models. It focuses on olympiad-level mathematics and complex multi-step mathematical reasoning.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
7 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.