2503.21380v2
Model: nemotron-free
# Challenging the Boundaries of Reasoning: An Olympiad-Level Math Benchmark for Large Language Models
**Authors**: DataCanvas Alaya NeW. BAAI.
Technical Report on Slow Thinking with LLMs: Evaluation Benchmark
## Abstract
In recent years, the rapid development of large reasoning models has resulted in the saturation of existing benchmarks for evaluating mathematical reasoning, highlighting the urgent need for more challenging and rigorous evaluation frameworks. To address this gap, we introduce OlymMATH, a novel Olympiad-level mathematical benchmark, designed to rigorously test the complex reasoning capabilities of LLMs. OlymMATH features 200 meticulously curated problems, each manually verified and available in parallel English and Chinese versions. The problems are systematically organized into two distinct difficulty tiers: (1) AIME-level problems (easy) that establish a baseline for mathematical reasoning assessment, and (2) significantly more challenging problems (hard) designed to push the boundaries of current state-of-the-art models. In our benchmark, these problems span four core mathematical fields, each including a verifiable numerical solution to enable objective, rule-based evaluation. Empirical results underscore the significant challenge presented by OlymMATH, with state-of-the-art models including DeepSeek-R1, OpenAIâs o3-mini and Gemini 2.5 Pro Exp demonstrating notably limited accuracy on the hard subset. Furthermore, the benchmark facilitates comprehensive bilingual assessment of mathematical reasoning abilitiesâa critical dimension that remains largely unaddressed in mainstream mathematical reasoning benchmarks. We release the benchmark, evaluation code, detailed results and a data visualization tool at https://github.com/RUCAIBox/OlymMATH.
## 1 Introduction
The advent of large language models (LLMs) [1] has marked a significant leap forward in the capabilities of artificial intelligence, showcasing exceptional performance across a broad spectrum of tasks, and in some cases, even rivaling or exceeding human-level proficiency [2, 3]. Among the myriad of capabilities demonstrated by LLMs, mathematical reasoning has surfaced as a particularly pivotal and demanding area of research [4, 5, 6]. In recent years, the evaluation and enhancement of mathematical reasoning abilities have become a central focus in the development of LLMs [7].
Effective assessment of LLM reasoning necessitates reliable and verifiable evaluation benchmarks. Reliability ensures accurately designed problems and solutions, free from ambiguities or errors. Verifiability demands that the evaluation process be easily constructed, replicated, and validated, often relying on easily parsable answer formats. Many benchmarks adopt a single-answer format, like â The answer is $3 3 3$ â, to simplify parsing and enhance reproducibility.
<details>
<summary>x1.png Details</summary>
### Visual Description
# Technical Document Extraction: AI Model Accuracy Comparison
## Chart Type
Bar chart comparing AI model accuracy across four mathematical problem datasets.
## Axes Labels
- **Y-axis**: Accuracy (%) [0-100 scale]
- **X-axis**: Datasets
- AIME 24 (30 Problems)
- HMMT 202502 (30 Problems)
- OlympMATH-EN-EASY (100 Problems)
- OlympMATH-EN-HARD (100 Problems)
## Legend
Right-aligned legend with 7 color-coded models:
1. đŁ Gemini 2.5 Pro Exp
2. đ© OpenAI o3-mini (high)
3. đš Qwen3-235B-A22B
4. đŠ Qwen3-30B-A3B
5. đ« DeepSeek-R1
6. đ« QwQ-32B
7. đ„ GLM-Z1-AIR
## Spatial Grounding
- Legend position: [x=100%, y=0-100%] (right edge)
- Dataset labels positioned at bottom center of each bar group
- Accuracy values displayed above each bar
## Data Points (Accuracy %)
### AIME 24 (30 Problems)
| Model | Accuracy |
|----------------------|----------|
| Gemini 2.5 Pro Exp | 92.0 |
| OpenAI o3-mini | 87.3 |
| Qwen3-235B-A22B | 85.7 |
| Qwen3-30B-A3B | 80.4 |
| DeepSeek-R1 | 79.8 |
| QwQ-32B | 79.5 |
| GLM-Z1-AIR | 80.8 |
### HMMT 202502 (30 Problems)
| Model | Accuracy |
|----------------------|----------|
| Gemini 2.5 Pro Exp | 82.5 |
| OpenAI o3-mini | 67.5 |
| Qwen3-235B-A22B | 62.5 |
| Qwen3-30B-A3B | 50.8 |
| DeepSeek-R1 | 41.7 |
| QwQ-32B | 47.5 |
| GLM-Z1-AIR | - |
### OlympMATH-EN-EASY (100 Problems)
| Model | Accuracy |
|----------------------|----------|
| Gemini 2.5 Pro Exp | 92.2 |
| OpenAI o3-mini | 91.4 |
| Qwen3-235B-A22B | 90.5 |
| Qwen3-30B-A3B | 87.2 |
| DeepSeek-R1 | 79.6 |
| QwQ-32B | 84.0 |
| GLM-Z1-AIR | 76.8 |
### OlympMATH-EN-HARD (100 Problems)
| Model | Accuracy |
|----------------------|----------|
| Gemini 2.5 Pro Exp | 58.4 |
| OpenAI o3-mini | 31.2 |
| Qwen3-235B-A22B | 36.5 |
| Qwen3-30B-A3B | 26.3 |
| DeepSeek-R1 | 19.5 |
| QwQ-32B | 23.1 |
| GLM-Z1-AIR | 20.1 |
## Key Trends
1. **Dataset Difficulty Gradient**:
- AIME 24 (easiest): Highest accuracies (79.5-92.0%)
- OlympMATH-EN-HARD (hardest): Lowest accuracies (19.5-58.4%)
2. **Model Performance Patterns**:
- Gemini 2.5 Pro Exp maintains top performance across all datasets
- OpenAI o3-mini shows strongest performance in OlympMATH-EN-EASY
- QwQ-32B demonstrates consistent mid-range performance
- DeepSeek-R1 shows significant performance drop in OlympMATH-EN-HARD
3. **Accuracy Degradation**:
- Average accuracy drop from AIME 24 to OlympMATH-EN-HARD:
- Gemini 2.5 Pro Exp: -33.6%
- OpenAI o3-mini: -56.1%
- Qwen3-235B-A22B: -49.2%
## Color Verification
All bar colors match legend specifications:
- đŁ = Gemini 2.5 Pro Exp (magenta)
- đ© = OpenAI o3-mini (green)
- đš = Qwen3-235B-A22B (yellow)
- đŠ = Qwen3-30B-A3B (blue)
- đ« = DeepSeek-R1 (gray)
- đ« = QwQ-32B (brown)
- đ„ = GLM-Z1-AIR (red)
## Data Table Reconstruction
| Dataset | Gemini 2.5 Pro Exp | OpenAI o3-mini | Qwen3-235B-A22B | Qwen3-30B-A3B | DeepSeek-R1 | QwQ-32B | GLM-Z1-AIR |
|-----------------------|--------------------|----------------|-----------------|---------------|-------------|---------|------------|
| AIME 24 (30) | 92.0 | 87.3 | 85.7 | 80.4 | 79.8 | 79.5 | 80.8 |
| HMMT 202502 (30) | 82.5 | 67.5 | 62.5 | 50.8 | 41.7 | 47.5 | - |
| OlympMATH-EN-EASY (100)| 92.2 | 91.4 | 90.5 | 87.2 | 79.6 | 84.0 | 76.8 |
| OlympMATH-EN-HARD (100)| 58.4 | 31.2 | 36.5 | 26.3 | 19.5 | 23.1 | 20.1 |
## Language Analysis
- All text in English
- No non-English content detected
</details>
Figure 1: Performance comparisons of mainstream reasoning models between our OlymMATH (English version) and other Olympiad-level mathematical benchmarks. Our OlymMATH dataset provides test results that align with those on existing benchmarks and features a significantly larger number of problems.
Evaluation benchmarks are primarily established to identify LLM limitations, and guiding future improvements. Over recent years, numerous high-quality mathematical benchmarks, such as GSM8K [8] and MATH [9], have been pivotal in advancing LLM reasoning capabilities [10, 11]. However, a significant trend is the saturation of many benchmarks, including those currently in use, due to rapid LLM advancements. For example, GSM8K [8], once a standard for earlier models like Llama 1 and 2, is now largely mastered by state-of-the-art models. Similarly, MATH [9], initially challenging for GPT-4-level models, has also become saturated by todayâs leading models. This saturation is further compounded by slow-thinking models like DeepSeek-R1 [4], OpenAIâs o3-mini [12], and Gemini 2.5 Pro Experimental [13]. These models, which promote deliberate step-by-step reasoning, show that enhancing the reasoning process yields substantial performance gains, thereby diminishing the effectiveness of existing benchmarks in differentiating cutting-edge capabilities.
To better evaluate the performance of advanced reasoning models, more rigorous and challenging benchmarks are needed to assess their mathematical reasoning capabilities. The AIME dataset has emerged as a more demanding benchmark by incorporating problems from the American Invitational Mathematics Examination (AIME), which presents a higher level of difficulty. Due to their complexity and rigor, AIME problems continue to challenge state-of-the-art models under standard prompting. Nevertheless, the AIME dataset has three major limitations. First, the limited scale of the current dataset (containing merely 30 problems from AIME 2024) may compromise the statistical reliability and robustness of the evaluation results. Second, as reasoning models rapidly improveâthrough methods like fine-tuning with long chain-of-thought data [14] or reinforcement learning scaling [4] âthe benchmarkâs original performance ceiling is being surpassed. For example, models such as Gemini 2.5 Pro Exp now achieve 92% accuracy with single attempt, demonstrating that current top-performing models are approaching the limits of what AIME can effectively measure. Third, the dataset exclusively features English problems, leaving multilingual reasoning capabilities unassessed despite their importance for a comprehensive evaluation.
To overcome these limitations, we present OlymMATH: a rigorously curated, bilingual (English and Chinese) benchmark for Olympiad-level reasoning, comprising 200 problems split into easy (OlymMATH-EASY) and hard (OlymMATH-HARD) levels with parallel bilingual sets (EN & ZH). To prevent data leakage, problems were manually sourced from printed publications and expert-verified. OlymMATH requires precise numerical answers for reliable verification, covers four major mathematical fields, and adheres to the MATH dataset [9] format for compatibility (see Figure 2).
MATH Dataset
Problem: Compute:
$$
1-2+3-4+5-\dots+99-100.
$$
Answer: $-50$ .
Problem: Let $n$ be a positive integer. Simplify the expression
$$
\frac{(2^{4}+\frac{1}{4})(4^{4}+\frac{1}{4})\dotsm[(2n)^{4}+\frac{1}{4}]}{(1^{
4}+\frac{1}{4})(3^{4}+\frac{1}{4})\dotsm[(2n-1)^{4}+\frac{1}{4}]}.
$$
Answer: $8n^{2}+4n+1$ .
v
OlymMATH-HARD (Ours) Problem-EN: UTF8gbsn Find the remainder of $\sum_{k=0}^{1234}\binom{2016\times 1234}{2016k}$ modulo $2017^{2}$ (provide the value in the range $[0,2017^{2})$ ). Answer: $1581330$ . Subject: Number Theory. OlymMATH-EASY (Ours) Problem-ZH: UTF8gbsn èźŸ $O$ äžș $\triangle ABC$ çć
ćż, $AB=3$ , $AC=4$ , $BC=5$ , $\overrightarrow{OP}=x\overrightarrow{OA}+y\overrightarrow{OB}+z\overrightarrow {OC}$ , $0\leqslant x,y,z\leqslant 1$ . æ±ćšçč $P$ ç蜚èżčæèŠççćčłéąćșćçéąç§Ż. Answer: $12$ . Subject: UTF8gbsnć äœ.
Figure 2: Examples from the MATH dataset and our OlymMATH dataset.
By leveraging the OlymMATH benchmark, we conduct extensive experiments to evaluate the performance of several state-of-the-art models (see Figure 1). The results underscore our benchmarkâs difficulty, with advanced models like DeepSeek-R1 [4], o3-mini [12], and Gemini 2.5 Pro Exp [13] achieving only 19.5%, 31.2%, and 58.4% accuracy, respectively, on OlymMATH-EN-HARD, indicating Olympiad-level math remains a significant challenge necessitating further research. Our multilingual comparison showed a consistent performance gap, with higher accuracy on English problems versus Chinese, highlighting the need for multilingual evaluation. Furthermore, case studies revealed models sometimes use heuristic âguessingâ to reach answers without rigorous proofs. This underscores the importance of process-level inspection for accurate LLM capability assessment.
In summary, our contribution are as follows:
$\bullet$ We introduce OlymMATH, a manually curated, Olympiad-level mathematical benchmark. It features parallel English and Chinese versions for objective, bilingual evaluation of LLM mathematical reasoning, with answers efficiently verifiable using sympy -based tools.
$\bullet$ Experiments demonstrate OlymMATHâs reliability (aligned with AIME) and strong discriminative power; even state-of-the-art models achieve only moderate scores, highlighting OlymMATHâs potential to drive LLM reasoning advancements.
$\bullet$ Detailed analyses and case studies reveal key model limitations in complex problem-solving, including performance disparities between English and Chinese problems and instances of heuristic âguessingâ rather than rigorous deduction.
$\bullet$ We open-source evaluation results and resources, including sampled long chain-of-thought reasoning trajectories (582,400 entries from 28 models on 400 problems), a data visualization tool, and standard solutions for problems where all LLMs struggled, to facilitate community research and analysis on diverse reasoning patterns and common reasoning issues.
## 2 Benchmark Construction
In this section, we describe the OlymMATH dataset in detail, including its construction methodology, problem composition, categorical distribution, and evaluation approach. Our dataset is specifically designed to provide a rigorous yet objectively verifiable benchmark for assessing the mathematical reasoning capabilities of LLMs. Additionally, we offer two parallel evaluation sets containing 200 problems each in English and Chinese as supplementary data to facilitate a comparative analysis of performance gaps between the two languages. Table 1 presents a basic comparison of our proposed OlymMATH benchmark and other mathematical reasoning benchmarks.
Table 1: Comparison of existing benchmarks. EN and ZH denote English and Chinese, respectively.
columneven = c, column3 = c, column5 = c, hline1,12 = -0.08em, hline2,10 = -0.05em, Name & # Problems # Field Language Evaluation Difficulty GSM8K [8] 1319 - EN Rule Grade School MATH [9] 5000 6 EN Rule Competition AIME 2024 [15] 30 - EN Rule Olympiad AIME 2025 [16] 30 - EN Rule Olympiad HMMT 202502 [17] 30 - EN Rule Olympiad USAMO 2025 [18] 6 - EN LLM Olympiad Olympiad Bench [19] 8476 3 ZH & EN Rule CEE & Olympiad Omni-MATH [20] 4428 33+ EN LLM Olympiad OlymMATH-EN 200 4 EN Rule Olympiad OlymMATH-ZH 200 4 ZH Rule Olympiad
### 2.1 Reliability: Contamination and Verification
#### Contamination
OlymMATH comprises 200 high-quality mathematical problems at the Olympiad level, meticulously curated from printed resources to ensure both quality and originality. These problems were manually gathered from a range of authoritative sources, including specialized magazines, textbooks, and official competition materials. To minimize the risk of data contamination, online repositories and forums were intentionally excluded from the sourcing process. This methodology ensures that the problems are intellectually challenging and representative of advanced mathematical reasoning, while also minimizing prior exposure on publicly accessible digital platforms. Consequently, OlymMATH serves as a reliable benchmark for evaluating the real capabilities of LLMs in solving complex mathematical tasks.
#### Verification
To enhance dataset reliability, we invited a China Mathematical Olympiad silver medalist and two provincial first-prize winners to verify and revise the problems and solutions. Since the answers to the problems were already provided, the verification difficulty was reduced, making the expertise of reviewers sufficient for this task. Each problem was reviewed by at least two reviewers. Additionally, official solutions for challenging problems are published for community oversight.
### 2.2 Problem Categories and Distribution
OlymMATH problems span four key high-school Olympiad mathematical fieldsâalgebra, geometry, number theory, and combinatoricsâclassified by experts (not LLMs) for reliability. Problems are selected for their challenge, suitability for simple-answer verification, and topic diversity (e.g., inequalities, sequences, and more in algebra). Figure-based problems within this set are text-reformulated for LLM compatibility, with non-convertible ones excluded (e.g., Figure 6 in Appendix).
For refined evaluation, problems are categorized by difficulty: easy, designed to challenge standard prompting in mainstream models, and hard, tailored to test advanced reasoning (e.g., slow-thinking modes) in state-of-the-art models. The distribution details are described in Table 2.
Table 2: The distribution of contest problems by category.
cells = c, cell61 = c=2, hline1,7 = -0.08em, hline2,6 = -0.05em, Category & Topic # HARD # EASY # Total Algebra (Alg.) Inequality, Sequence, Trigonometry, etc. 25 25 50 Geometry (Geo.) Solid Geometry, Analytic Geometry, etc. 25 33 58 Number Theory (Num.) Divisibility, Diophantine Equation, etc. 25 13 38 Combinatorics (Com.) Graph Theory, Permutation, etc. 25 29 54 Total 100 100 200
### 2.3 Format and Verification Methodology
OlymMATH adopts the MATH dataset format (see Figure 2) for seamless integration with existing pipelines and enhancing clarity and processing efficiency. All problems are text-based, including geometry reformulated from diagrams to align with LLM evaluation, as mentioned previously. For consistent, objective assessment, answers are restricted to real numbers and intervals (see Table 3), avoiding ambiguous formats and enabling reliable sympy -based and numerical verification.
Table 3: The included and excluded formats of the final answer.
width = colspec = Q[c,m,wd=0.16] X[l,m] X[l,m], cell12 = halign=c, cell13 = halign=c, cell31 = r=2c, hline1,Z = 0.08em, solid, hline2 = 0.05em, solid, hline3 = 0.05em, solid & Type & Example Type & Example Included Real number: $16^{\circ}$ , $2^{2017}+\arctan 2$ Interval: $[\sqrt{33},+\infty)$ , $(4,5\pi]$ Excluded Set Operations: $\{4,5\}\cup\{1,8\}$ Variable: $\sqrt[3]{5}a^{2}$ , $p^{2}-pq$ , $n!+2$ Complex number: $9+4\mathrm{i}$ , $\sqrt{-4}$ Text: East, Alice
To make the evaluation more challenging, OlymMATH includes problems with multiple numerical answers. These problems are modified to require a summary of all potential outcomes (e.g., sums, sums of squares; see Figure 7 in Appendix). This method effectively assesses whether models can consider all possible answers, thereby providing a robust evaluation of their reasoning capabilities.
### 2.4 Bilingual Extension
Originating from Chinese-language problems, the OlymMATH benchmark includes both original Chinese and translated English versions for comprehensive bilingual evaluation. Our LLM-based translation pipeline first uses Claude Sonnet 3.7 for initial English translations, which are then iteratively refined with GPT-4o. Finally, a crucial human verification stage by two expert annotators ensures mathematical accuracy, rigor, and linguistic fluency. These resulting parallel sets, OlymMATH-EN (English) and OlymMATH-ZH (Chinese) (see Figure 2), facilitate systematic comparison of cross-lingual reasoning, with their union denoted as OlymMATH (full set).
## 3 Experiments
In this section, we assess the performance of leading reasoning models using the OlymMATH benchmark and then provide a detailed analysis of their capabilities.
### 3.1 Experimental Setup
#### Models.
To conduct a thorough evaluation, we assess a range of representative LLMs. For open-source models, we investigated recent work on reasoning models, and evaluated DeepSeek-R1 series [4], STILL-3-Preview [21], DeepScaleR-Preview [22], QwQ [23], Light-R1 series [24], OpenThinker2 series [25], Skywork-OR1 series [26], GLM-Z1-Air [27], AceMath-RL [28], OpenMath-Nemotron series [29], and Qwen3 series [30]. For closed-source models, we include o3-mini (high) [12], Gemini 2.5 Pro Experimental 0325 [13] in our evaluation.
Table 4: Model performance on OlymMATH-EN. Models within each model size group are sorted by release time. The abbreviations âAlg.â, âGeo.â, âNum.â, and âCom.â represent the four categories in OlymMATH. Highest accuracy per model size is bolded. The second highest accuracy per model size is underlined. Models sampled only 8 times are marked in gray to indicate potential instability.
| Model | OlymMATH-EN-HARD | OlymMATH-EN-EASY | | | | | | | | | | | | | | | | | | |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Alg. | Geo. | Num. | Com. | Avg. | Alg. | Geo. | Num. | Com. | Avg. | | | | | | | | | | | |
| P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | |
| Qwen3 (0.6B, Think) | 2.5 | 0.0 | 2.1 | 4.0 | 6.6 | 8.0 | 0.2 | 0.0 | 2.8 | 3.0 | 15.5 | 20.0 | 5.6 | 15.2 | 24.5 | 38.5 | 5.2 | 6.9 | 10.4 | 17.0 |
| DS-R1-Distill (1.5B) | 1.9 | 0.0 | 1.8 | 0.0 | 1.8 | 0.0 | 0.4 | 0.0 | 1.5 | 0.0 | 20.8 | 40.0 | 12.6 | 21.2 | 32.6 | 61.5 | 8.2 | 24.1 | 16.0 | 32.0 |
| STILL-3-Pre. (1.5B) | 3.7 | 0.0 | 4.9 | 4.0 | 5.8 | 8.0 | 0.8 | 0.0 | 3.8 | 3.0 | 22.7 | 36.0 | 14.8 | 30.3 | 37.6 | 69.2 | 10.3 | 17.2 | 18.4 | 33.0 |
| DeepScaleR-Pre. (1.5B) | 3.4 | 4.0 | 4.2 | 8.0 | 8.2 | 4.0 | 0.4 | 0.0 | 4.1 | 4.0 | 19.9 | 16.0 | 18.5 | 21.2 | 44.6 | 46.2 | 18.9 | 31.0 | 22.3 | 26.0 |
| OpenMath-Nemo. (1.5B) | 14.5 | 24.0 | 13.6 | 16.0 | 10.9 | 16.0 | 2.6 | 4.0 | 10.4 | 15.0 | 70.9 | 100.0 | 59.3 | 90.9 | 81.6 | 100.0 | 40.6 | 58.6 | 59.7 | 85.0 |
| Qwen3 (4B, Think) | 18.1 | 20.0 | 14.8 | 12.0 | 19.8 | 28.0 | 3.1 | 4.0 | 13.9 | 16.0 | 76.4 | 92.0 | 79.1 | 97.0 | 85.1 | 84.6 | 57.1 | 72.4 | 72.8 | 87.0 |
| DS-R1-Distill (7B) | 15.6 | 36.0 | 12.6 | 24.0 | 13.1 | 24.0 | 3.1 | 4.0 | 11.1 | 22.0 | 52.8 | 84.0 | 49.6 | 84.8 | 62.5 | 84.6 | 33.9 | 58.6 | 47.5 | 77.0 |
| Light-R1-DS (7B) | 17.1 | 28.0 | 15.2 | 16.0 | 12.8 | 24.0 | 3.6 | 4.0 | 12.2 | 18.0 | 57.1 | 84.0 | 53.6 | 93.9 | 73.7 | 84.6 | 39.5 | 51.7 | 53.0 | 78.0 |
| OpenThinker2 (7B) | 16.0 | 20.0 | 16.8 | 28.0 | 14.0 | 20.0 | 2.8 | 4.0 | 12.4 | 18.0 | 65.3 | 96.0 | 60.5 | 97.0 | 79.1 | 84.6 | 42.3 | 58.6 | 58.9 | 84.0 |
| Skywork-OR1-Pre. (7B) | 14.4 | 20.0 | 12.5 | 12.0 | 11.7 | 24.0 | 1.6 | 0.0 | 10.0 | 14.0 | 61.6 | 88.0 | 55.9 | 78.8 | 74.3 | 92.3 | 36.9 | 48.3 | 54.2 | 74.0 |
| Skywork-OR1-Math (7B) | 17.4 | 20.0 | 17.1 | 20.0 | 13.6 | 28.0 | 0.9 | 0.0 | 12.2 | 17.0 | 67.9 | 92.0 | 67.4 | 93.9 | 76.6 | 92.3 | 47.6 | 62.1 | 63.0 | 84.0 |
| AceMath-RL (7B) | 19.4 | 32.0 | 19.3 | 32.0 | 14.4 | 24.0 | 3.5 | 4.0 | 14.2 | 23.0 | 69.7 | 96.0 | 63.7 | 93.9 | 79.0 | 84.6 | 44.2 | 69.0 | 61.5 | 86.0 |
| OpenMath-Nemo. (7B) | 26.9 | 36.0 | 18.6 | 28.0 | 19.8 | 28.0 | 4.4 | 4.0 | 17.4 | 24.0 | 86.4 | 100.0 | 76.4 | 97.0 | 91.5 | 100.0 | 55.3 | 72.4 | 74.7 | 91.0 |
| DS-R1-Distill (14B) | 16.1 | 16.0 | 17.0 | 16.0 | 18.1 | 32.0 | 2.1 | 4.0 | 13.3 | 17.0 | 69.0 | 96.0 | 65.1 | 97.0 | 79.4 | 92.3 | 44.0 | 65.5 | 61.8 | 87.0 |
| Light-R1-DS (14B) | 21.8 | 24.0 | 22.2 | 28.0 | 17.8 | 36.0 | 2.6 | 4.0 | 16.1 | 23.0 | 72.3 | 88.0 | 73.0 | 100.0 | 84.3 | 92.3 | 47.6 | 65.5 | 66.9 | 86.0 |
| OpenMath-Nemo. (14B) | 28.7 | 40.0 | 22.1 | 32.0 | 21.0 | 32.0 | 3.4 | 4.0 | 18.8 | 27.0 | 87.9 | 100.0 | 78.5 | 93.9 | 95.8 | 100.0 | 59.9 | 86.2 | 77.7 | 94.0 |
| Qwen3 (30B-A3B, Think) | 38.8 | 44.0 | 33.8 | 44.0 | 26.7 | 36.0 | 5.9 | 4.0 | 26.3 | 32.0 | 91.4 | 100.0 | 92.9 | 100.0 | 90.9 | 92.3 | 75.6 | 93.1 | 87.2 | 97.0 |
| DS-R1-Distill (32B) | 22.4 | 32.0 | 21.4 | 24.0 | 20.3 | 40.0 | 3.4 | 4.0 | 16.9 | 25.0 | 73.6 | 100.0 | 71.8 | 97.0 | 84.5 | 92.3 | 49.0 | 69.0 | 67.3 | 89.0 |
| QwQ (32B) | 32.9 | 28.0 | 26.6 | 36.0 | 26.7 | 44.0 | 6.2 | 4.0 | 23.1 | 28.0 | 91.8 | 100.0 | 87.0 | 100.0 | 95.0 | 100.0 | 69.0 | 89.7 | 84.0 | 97.0 |
| Light-R1-DS (32B) | 28.9 | 44.0 | 31.1 | 52.0 | 24.1 | 36.0 | 5.2 | 8.0 | 22.3 | 35.0 | 84.2 | 100.0 | 83.3 | 100.0 | 92.5 | 100.0 | 62.1 | 82.8 | 78.6 | 95.0 |
| OpenThinker2 (32B) | 24.1 | 32.0 | 22.9 | 32.0 | 18.0 | 20.0 | 2.6 | 4.0 | 16.9 | 22.0 | 79.4 | 96.0 | 74.0 | 100.0 | 90.4 | 92.3 | 56.5 | 79.3 | 72.4 | 92.0 |
| Skywork-OR1-Pre. (32B) | 37.2 | 52.0 | 32.3 | 48.0 | 27.0 | 40.0 | 4.2 | 4.0 | 25.2 | 36.0 | 89.3 | 100.0 | 87.3 | 100.0 | 92.4 | 100.0 | 63.9 | 82.8 | 81.7 | 95.0 |
| GLM-Z1-Air (32B) | 35.0 | 44.0 | 21.5 | 32.0 | 19.5 | 24.0 | 4.5 | 4.0 | 20.1 | 26.0 | 86.5 | 100.0 | 79.5 | 90.9 | 90.4 | 100.0 | 59.1 | 75.9 | 76.8 | 90.0 |
| OpenMath-Nemo. (32B) | 22.0 | 36.0 | 21.0 | 28.0 | 20.0 | 24.0 | 3.5 | 4.0 | 16.6 | 23.0 | 75.5 | 100.0 | 60.6 | 90.9 | 89.4 | 100.0 | 42.2 | 69.0 | 62.7 | 88.0 |
| Qwen3 (235B-A22B, Think) | 48.0 | 52.0 | 49.5 | 60.0 | 38.0 | 36.0 | 10.5 | 16.0 | 36.5 | 41.0 | 93.5 | 100.0 | 92.4 | 100.0 | 99.0 | 100.0 | 81.9 | 93.1 | 90.5 | 98.0 |
| DeepSeek R1 | 30.0 | 40.0 | 25.5 | 32.0 | 18.5 | 24.0 | 4.0 | 4.0 | 19.5 | 25.0 | 90.5 | 100.0 | 82.2 | 97.0 | 94.2 | 100.0 | 60.8 | 72.4 | 79.6 | 91.0 |
| OpenAI o3-mini (high) | 29.5 | 32.0 | 29.0 | 44.0 | 49.5 | 60.0 | 17.0 | 20.0 | 31.2 | 39.0 | 93.0 | 92.0 | 89.8 | 100.0 | 97.1 | 100.0 | 89.2 | 96.6 | 91.4 | 97.0 |
| Gemini 2.5 Pro Exp 0325 | 71.5 | 76.0 | 75.5 | 84.0 | 59.0 | 72.0 | 27.5 | 36.0 | 58.4 | 67.0 | 92.0 | 100.0 | 97.0 | 100.0 | 98.1 | 100.0 | 84.5 | 89.7 | 92.2 | 97.0 |
#### Evaluation Details.
Our evaluation pipeline follows a systematic approach: for each problem, we generate 64 distinct responses from each comparison model, with the exception of certain models (i.e., OpenMath-Nemotron-32B, Qwen3-235B-A22B, GLM-Z1-Air, DeepSeek-R1, o3-mini (high) and Gemini 2.5 Pro Exp), for which, due to resource limitations and the relatively large scale of our dataset, we only conducted 8 samples. For the Pass@1 metric, we compute the mean accuracy across all sampled responses to derive the final accuracy score. For the Cons@64 and Cons@8 metric, we implement majority voting to determine a consensus answer for each problem, subsequently calculating the average accuracy across the entire dataset. For generation hyperparameters, we adhere to established practices from previous research [4, 23], configuring locally-evaluated models with temperature, top_p, min_p, and max_token set to $0.6$ , $0.95$ , $0 0$ , and $32768$ , respectively. For api-evaluated models (i.e., GLM-Z1-Air, DeepSeek-R1, o3-mini (high) and Gemini 2.5 Pro Exp), we expand their max_token limit to the maximum extent possible to unleash their reasoning capabilities better. We have open-sourced all the samples (a dataset of 582,400 math reasoning samples with long chain-of-thought, generated from the 400 problems in OlymMATH across 28 models), an online data visualization tool and standard solutions for problems where all LLMs struggled in our repository, aiming to help the community analyze the problem-solving patterns and characteristics of LLMs (see Section 4 for further information).
### 3.2 Evaluation Results
In this part, we assess the performance of reasoning models on our benchmark. We present the evaluation results of OlymMATH-EN and OlymMATH-ZH in Table 4 and Table 5, respectively.
Table 5: Model performance on OlymMATH-ZH. Models within each model size group are sorted by release time. The abbreviations âAlg.â, âGeo.â, âNum.â, and âCom.â represent the four categories in OlymMATH. Highest accuracy per model size is bolded. The second highest accuracy per model size is underlined. Models sampled only 8 times are marked in gray to indicate potential instability.
| Model | OlymMATH-ZH-HARD | OlymMATH-ZH-EASY | | | | | | | | | | | | | | | | | | |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Alg. | Geo. | Num. | Com. | Avg. | Alg. | Geo. | Num. | Com. | Avg. | | | | | | | | | | | |
| P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | P@1 | C@k | |
| Qwen3 (0.6B, Think) | 2.6 | 4.0 | 0.8 | 0.0 | 4.4 | 4.0 | 0.0 | 0.0 | 1.9 | 2.0 | 9.9 | 8.0 | 2.8 | 3.0 | 12.0 | 15.4 | 1.3 | 3.4 | 5.4 | 6.0 |
| DS-R1-Distill (1.5B) | 1.8 | 0.0 | 1.3 | 0.0 | 1.1 | 0.0 | 0.0 | 0.0 | 1.0 | 0.0 | 13.7 | 20.0 | 6.3 | 9.1 | 20.9 | 30.8 | 2.6 | 0.0 | 9.0 | 12.0 |
| STILL-3-Pre. (1.5B) | 2.9 | 0.0 | 2.2 | 0.0 | 4.5 | 4.0 | 0.2 | 0.0 | 2.5 | 1.0 | 15.9 | 32.0 | 7.4 | 18.2 | 27.6 | 46.2 | 4.3 | 6.9 | 11.3 | 22.0 |
| DeepScaleR-Pre. (1.5B) | 4.4 | 8.0 | 2.6 | 4.0 | 6.4 | 8.0 | 0.1 | 0.0 | 3.4 | 5.0 | 15.9 | 20.0 | 7.2 | 9.1 | 32.6 | 46.2 | 8.9 | 20.7 | 13.2 | 20.0 |
| OpenMath-Nemo. (1.5B) | 13.9 | 16.0 | 9.8 | 4.0 | 13.3 | 16.0 | 0.8 | 0.0 | 9.5 | 9.0 | 67.9 | 96.0 | 37.6 | 57.6 | 65.3 | 76.9 | 27.6 | 41.4 | 45.9 | 65.0 |
| Qwen3 (4B, Think) | 12.5 | 20.0 | 7.0 | 8.0 | 12.6 | 24.0 | 0.9 | 0.0 | 8.3 | 13.0 | 70.8 | 88.0 | 61.0 | 75.8 | 74.8 | 92.3 | 41.8 | 51.7 | 59.7 | 74.0 |
| DS-R1-Distill (7B) | 6.1 | 8.0 | 7.9 | 12.0 | 6.6 | 8.0 | 0.6 | 0.0 | 5.3 | 7.0 | 38.0 | 64.0 | 30.8 | 51.5 | 49.2 | 61.5 | 18.7 | 27.6 | 31.5 | 49.0 |
| Light-R1-DS (7B) | 7.1 | 4.0 | 9.4 | 12.0 | 7.8 | 12.0 | 1.1 | 0.0 | 6.3 | 7.0 | 42.9 | 76.0 | 42.7 | 72.7 | 56.9 | 61.5 | 22.7 | 31.0 | 38.8 | 60.0 |
| OpenThinker2 (7B) | 7.0 | 0.0 | 7.3 | 8.0 | 7.4 | 8.0 | 1.0 | 0.0 | 5.7 | 4.0 | 48.2 | 80.0 | 44.7 | 72.7 | 57.8 | 76.9 | 22.4 | 37.9 | 40.8 | 65.0 |
| Skywork-OR1-Pre. (7B) | 4.7 | 4.0 | 7.8 | 8.0 | 7.4 | 8.0 | 0.4 | 0.0 | 5.1 | 5.0 | 41.1 | 60.0 | 36.6 | 54.5 | 58.1 | 69.2 | 23.6 | 34.5 | 36.8 | 52.0 |
| Skywork-OR1-Math (7B) | 6.4 | 8.0 | 8.3 | 8.0 | 9.8 | 12.0 | 0.8 | 0.0 | 6.3 | 7.0 | 45.2 | 72.0 | 40.0 | 63.6 | 62.3 | 69.2 | 30.2 | 37.9 | 41.3 | 59.0 |
| AceMath-RL (7B) | 6.4 | 8.0 | 10.7 | 12.0 | 7.8 | 8.0 | 1.4 | 0.0 | 6.6 | 7.0 | 55.1 | 88.0 | 46.6 | 75.8 | 66.9 | 76.9 | 31.0 | 44.8 | 46.9 | 70.0 |
| OpenMath-Nemo. (7B) | 25.0 | 32.0 | 20.8 | 28.0 | 22.3 | 36.0 | 4.8 | 4.0 | 18.2 | 25.0 | 86.8 | 100.0 | 72.7 | 90.9 | 91.8 | 100.0 | 57.9 | 79.3 | 74.4 | 91.0 |
| DS-R1-Distill (14B) | 5.2 | 0.0 | 5.3 | 4.0 | 8.7 | 16.0 | 0.2 | 0.0 | 4.9 | 5.0 | 43.1 | 56.0 | 38.9 | 66.7 | 58.2 | 69.2 | 24.8 | 31.0 | 38.4 | 54.0 |
| Light-R1-DS (14B) | 6.2 | 4.0 | 7.5 | 8.0 | 10.9 | 12.0 | 0.2 | 0.0 | 6.2 | 6.0 | 56.6 | 84.0 | 45.5 | 75.8 | 66.5 | 76.9 | 28.7 | 37.9 | 46.1 | 67.0 |
| OpenMath-Nemo. (14B) | 28.7 | 32.0 | 26.1 | 40.0 | 26.8 | 40.0 | 4.2 | 4.0 | 21.4 | 29.0 | 88.3 | 100.0 | 75.2 | 100.0 | 94.5 | 100.0 | 60.2 | 86.2 | 76.6 | 96.0 |
| Qwen3 (30B-A3B, Think) | 35.6 | 40.0 | 24.1 | 28.0 | 18.1 | 24.0 | 2.7 | 4.0 | 20.1 | 24.0 | 87.8 | 92.0 | 84.7 | 97.0 | 91.3 | 100.0 | 61.9 | 65.5 | 79.7 | 87.0 |
| DS-R1-Distill (32B) | 6.5 | 0.0 | 5.4 | 4.0 | 10.6 | 12.0 | 0.7 | 0.0 | 5.8 | 4.0 | 45.2 | 52.0 | 41.8 | 63.6 | 60.2 | 69.2 | 26.0 | 37.9 | 40.4 | 54.0 |
| QwQ (32B) | 20.9 | 24.0 | 15.9 | 16.0 | 17.6 | 24.0 | 2.0 | 0.0 | 14.1 | 16.0 | 85.4 | 96.0 | 76.6 | 97.0 | 92.9 | 100.0 | 53.8 | 69.0 | 74.3 | 89.0 |
| Light-R1-DS (32B) | 16.8 | 28.0 | 12.0 | 12.0 | 13.4 | 16.0 | 4.4 | 16.0 | 11.6 | 18.0 | 70.1 | 96.0 | 64.1 | 93.9 | 80.4 | 92.3 | 39.8 | 51.7 | 60.7 | 82.0 |
| OpenThinker2 (32B) | 13.6 | 16.0 | 11.1 | 16.0 | 12.7 | 20.0 | 0.9 | 0.0 | 9.6 | 13.0 | 68.0 | 92.0 | 64.3 | 93.9 | 84.6 | 92.3 | 44.8 | 65.5 | 62.2 | 85.0 |
| Skywork-OR1-Pre. (32B) | 19.6 | 20.0 | 16.8 | 20.0 | 18.9 | 24.0 | 3.5 | 4.0 | 14.7 | 17.0 | 79.5 | 96.0 | 72.1 | 93.9 | 88.0 | 100.0 | 45.4 | 58.6 | 68.3 | 85.0 |
| GLM-Z1-Air (32B) | 18.0 | 16.0 | 12.0 | 8.0 | 16.0 | 16.0 | 2.5 | 4.0 | 12.1 | 11.0 | 76.0 | 96.0 | 69.3 | 78.8 | 89.4 | 92.3 | 41.8 | 48.3 | 65.6 | 76.0 |
| OpenMath-Nemo. (32B) | 22.5 | 36.0 | 22.5 | 32.0 | 22.5 | 28.0 | 3.5 | 4.0 | 17.8 | 25.0 | 68.0 | 96.0 | 62.5 | 90.9 | 90.4 | 100.0 | 48.7 | 72.4 | 63.5 | 88.0 |
| Qwen3 (235B-A22B, Think) | 36.5 | 48.0 | 43.5 | 48.0 | 28.5 | 32.0 | 4.0 | 8.0 | 28.1 | 34.0 | 91.0 | 100.0 | 90.2 | 97.0 | 94.2 | 100.0 | 78.4 | 89.7 | 87.5 | 96.0 |
| DeepSeek R1 | 20.0 | 24.0 | 25.0 | 28.0 | 17.0 | 16.0 | 1.5 | 0.0 | 15.9 | 17.0 | 79.5 | 96.0 | 74.6 | 84.8 | 88.5 | 92.3 | 49.6 | 55.2 | 70.4 | 80.0 |
| OpenAI o3-mini (high) | 31.5 | 40.0 | 32.5 | 44.0 | 48.5 | 56.0 | 19.0 | 28.0 | 32.9 | 42.0 | 93.0 | 96.0 | 89.4 | 100.0 | 99.0 | 100.0 | 85.8 | 93.1 | 90.5 | 97.0 |
| Gemini 2.5 Pro Exp 0325 | 65.0 | 76.0 | 78.0 | 80.0 | 53.5 | 56.0 | 25.0 | 40.0 | 55.4 | 63.0 | 90.5 | 96.0 | 93.2 | 93.9 | 100.0 | 100.0 | 84.1 | 86.2 | 90.8 | 93.0 |
First, we observe that all tested models exhibit relatively poor performance, with even OpenAI o3-mini (high) and Gemini 2.5 Pro Exp achieving only 31.2% and 58.4% on OlymMATH-EN-HARD. This underscores the high overall difficulty of our benchmark, which demands stronger reasoning abilities and a deeper understanding of mathematical knowledge to solve the problems effectively. In contrast, the performance of these advanced reasoning models on OlymMATH-EN-EASY is more modest and comparable to that on AIME 2024 and AIME 2025, suggesting that OlymMATH-EN-EASY is well-suited for evaluating the capabilities of less advanced reasoning models.
Second, by comparing the performance of LLMs on OlymMATH-EN and OlymMATH-ZH, we find that language can influence the reasoning performance of LLMs to some extent (see Figure 3). Overall, all models tend to achieve higher performance on the English benchmarks. A potential reason for this is that English corpora still dominate existing pre-training datasets, making the English-based task-solving capabilities of LLMs generally more superior compared to other languages. This finding highlights the importance of considering performance across different languages when conducting a comprehensive evaluation of LLMs.
<details>
<summary>x2.png Details</summary>
### Visual Description
# Technical Document Extraction: Scatter Plot Analysis
## Image Description
The image is a **scatter plot** with **two subplots** (left and right) comparing performance metrics of various AI models. The plots use **color-coded data points** to represent model parameters (in billions) and include **labels for specific models**. A **dashed identity line** (y = x) is present in both subplots for reference.
---
## Key Components
### 1. **Axes Labels**
- **Left Subplot (EASY):**
- **X-axis:** `OLYMATH ZH-EASY (pass@1)`
- **Y-axis:** `OLYMATH EN-EASY (pass@1)`
- **Right Subplot (HARD):**
- **X-axis:** `OLYMATH ZH-HARD (pass@1)`
- **Y-axis:** `OLYMATH EN-HARD (pass@1)`
### 2. **Legend**
- **Color Bar:**
- **Title:** `Model Parameters (Billions)`
- **Range:** `1.5B` (purple) to `32.0B` (yellow)
- **Position:** Right side of the image, vertical orientation.
### 3. **Data Points & Labels**
- **Left Subplot (EASY):**
- Data points are clustered near the identity line, with larger models (yellow) achieving higher pass@1 scores.
- **Labels (from top to bottom):**
- `Gemini 2.5 Pro Exp`
- `o3-mini (high)`
- `Qwen-335B-A22B`
- `Qwen3-30B-A3B`
- `Skywork-OR1-32B`
- `Qwen-32B`
- `Qwen3-35B-A22B`
- `Light-R1-DS-32B`
- `GLM-Z1-AIR`
- `DeepSeek-R1`
- `OpenMath-7B`
- `OpenMath-14B`
- `OpenMath-32B`
- `OpenMath-1.5B`
- `Skywork-OR1-7B`
- `OpenThinker2-7B`
- `DS-R1-7B`
- `Light-R1-DS-7B`
- `Skywork-OR1-32B`
- `DS-R1-14B`
- `AceMath-RL-7B`
- `OpenMath-32B`
- `DS-R1-32B`
- `Skywork-OR1-7B`
- `DS-R1-1.5B`
- `STILL-3-1.5B`
- `DeepScaler-1.5B`
- **Right Subplot (HARD):**
- Data points are more dispersed, with some models (e.g., `Gemini 2.5 Pro Exp`) achieving higher pass@1 scores than their ZH-HARD counterparts.
- **Labels (from top to bottom):**
- `Gemini 2.5 Pro Exp`
- `o3-mini (high)`
- `Qwen3-235B-A22B`
- `Qwen3-30B-A3B`
- `Skywork-OR1-32B`
- `Qwen-32B`
- `Qwen3-35B-A22B`
- `Light-R1-DS-32B`
- `GLM-Z1-AIR`
- `DeepSeek-R1`
- `OpenMath-7B`
- `OpenMath-14B`
- `OpenMath-32B`
- `OpenMath-1.5B`
- `Skywork-OR1-7B`
- `OpenThinker2-7B`
- `DS-R1-7B`
- `Light-R1-DS-7B`
- `Skywork-OR1-32B`
- `DS-R1-14B`
- `AceMath-RL-7B`
- `OpenMath-32B`
- `DS-R1-32B`
- `Skywork-OR1-7B`
- `DS-R1-1.5B`
- `STILL-3-1.5B`
- `DeepScaler-1.5B`
### 4. **Dashed Identity Line**
- A **dashed line** (y = x) is present in both subplots, serving as a reference for comparing ZH and EN performance.
---
## Spatial Grounding
- **Legend Position:** Right side of the image, vertical color bar.
- **Color Matching:**
- `Gemini 2.5 Pro Exp` (yellow) corresponds to `32.0B` in the legend.
- `o3-mini (high)` (yellow) corresponds to `32.0B`.
- `Qwen3-30B-A3B` (green) corresponds to `24.4B`.
- `DS-R1-1.5B` (purple) corresponds to `1.5B`.
---
## Trend Verification
- **Left Subplot (EASY):**
- Data points generally follow the identity line, with larger models (yellow) achieving higher pass@1 scores.
- Example: `Gemini 2.5 Pro Exp` (32.0B) has a pass@1 of ~0.85 on both axes.
- **Right Subplot (HARD):**
- Data points are more spread out, with some models (e.g., `Gemini 2.5 Pro Exp`) achieving higher pass@1 scores on the EN-HARD axis than on the ZH-HARD axis.
- Example: `Gemini 2.5 Pro Exp` (32.0B) has a pass@1 of ~0.6 on the EN-HARD axis and ~0.55 on the ZH-HARD axis.
---
## Notes
- **Language:** All text is in **English**.
- **Data Table:** No explicit data table is present; data is represented via scatter points and labels.
- **Missing Information:** No additional textual or numerical data is embedded in the image.
---
## Conclusion
The image compares the performance of AI models on **ZH-EASY**, **EN-EASY**, **ZH-HARD**, and **EN-HARD** tasks, with model parameters (in billions) represented via color. Larger models (e.g., `Gemini 2.5 Pro Exp`) generally perform better, but performance varies between ZH and EN tasks.
</details>
Figure 3: Pass@1 accuracy on OlymMATH EN (y) vs. ZH (x), the dashed line shows parity. Points above favor English, below favor Chinese. Solid circles (local dense models, colored by size) indicate larger models trend towards higher accuracy. Hollow diamonds are MoE or API evaluated models.
### 3.3 Benchmark Comparison
<details>
<summary>x3.png Details</summary>
### Visual Description
# Technical Document Extraction: AI/ML Model Performance Analysis
## Chart Type
Scatter plot comparing AI/ML model performance across two evaluation datasets.
## Axes
- **X-axis**: "AI/ML Accuracy" (range: 0.1 to 1.0)
- **Y-axis**: "OlymMATH EN Accuracy (pass@1)" (range: 0.0 to 1.0)
## Legend
- **Location**: Top-right corner
- **Components**:
- `EN-EASY`: Teal circles (solid line trend)
- `EN-HARD`: Green triangles (dashed line trend)
## Data Series
### EN-EASY (Teal Circles)
- **Trend**: Upward-sloping dashed blue line
- **Key Data Points**:
- [0.3, 0.15] - Unlabeled
- [0.45, 0.22] - Unlabeled
- [0.5, 0.58] - "OpenMATH-1.5B"
- [0.6, 0.52] - "OpenMATH-14B"
- [0.65, 0.56] - "Qwen3-4B"
- [0.7, 0.65] - "Qwen3-30B-A3B"
- [0.75, 0.72] - "Qwen3-235B-A22B"
- [0.8, 0.78] - "Gemini 2.5 Pro Exp"
- [0.85, 0.88] - "o3-mini (high)"
### EN-HARD (Green Triangles)
- **Trend**: Upward-sloping dashed red line
- **Key Data Points**:
- [0.3, 0.02] - Unlabeled
- [0.45, 0.05] - Unlabeled
- [0.5, 0.58] - "OpenMATH-1.5B"
- [0.6, 0.52] - "OpenMATH-14B"
- [0.65, 0.56] - "Qwen3-4B"
- [0.7, 0.65] - "Qwen3-30B-A3B"
- [0.75, 0.72] - "Qwen3-235B-A22B"
- [0.8, 0.78] - "Gemini 2.5 Pro Exp"
- [0.85, 0.88] - "o3-mini (high)"
## Color Gradient
- **Bar**: Right-side vertical color bar
- **Range**: January 2025 (dark purple) â April 2025 (bright yellow)
- **Mapping**: Data point color intensity corresponds to release date
## Annotations
- **Special Labels**:
- "o3-mini (high)" appears twice (EN-EASY and EN-HARD)
- "Gemini 2.5 Pro Exp" appears twice (EN-EASY and EN-HARD)
## Spatial Grounding
- **Legend**: Top-right corner (coordinates: [0.85, 0.95] to [0.95, 1.0])
- **Color Bar**: Right edge (coordinates: [1.0, 0.0] to [1.0, 1.0])
## Trend Verification
1. **EN-EASY Trend**:
- Slope: ~0.7 (y = 0.7x - 0.05)
- Correlation: Strong positive relationship
2. **EN-HARD Trend**:
- Slope: ~0.5 (y = 0.5x - 0.03)
- Correlation: Moderate positive relationship
## Component Isolation
1. **Header**: Chart title (not explicitly visible)
2. **Main Chart**:
- Scatter plot with two data series
- Two trend lines
3. **Footer**: Color bar and legend
## Critical Observations
1. EN-EASY models consistently outperform EN-HARD models
2. Later release dates (yellow) show higher accuracy
3. "o3-mini (high)" achieves near-perfect scores in both datasets
4. EN-HARD trend shows diminishing returns at higher accuracy levels
## Missing Information
- No explicit chart title
- No grid lines visible
- No data table present
- No secondary y-axis
## Language Notes
- All text in English
- No non-English content detected
</details>
Figure 4: Correlation of Pass@1 performance: OlymMATH-EN vs. AIME24. Dashed lines indicate linear trends per dataset. Solid shapes are local dense models (size = model size, color = release date). Hollow shapes denote MoE or API evaluated models. Stars mark the best overall model.
To comprehensively evaluate OlymMATH against existing benchmarks, we compare state-of-the-art model performance across widely used mathematical benchmarks (see Figure 1). Results are sourced from research reports or the MathArena platform https://matharena.ai/ .
Figure 1 illustrates that OlymMATH is more challenging, yielding lower accuracy compared to saturated benchmarks like MATH-500 (where even DeepSeek-R1-Distill-Qwen-7B exceeds 92% accuracy [4]) or AIME24 (where top LLMs reach 92% with single attempt). Unlike these benchmarks whose high performance limits discriminative power, OlymMATH elicits more varied scores, offering superior differentiation of reasoning capabilities. For example, while Gemini 2.5 Pro Exp and o3-mini (high) achieve similar AIME24 accuracy (92.0% vs. 87.3%), their OlymMATH-EN-HARD performance diverges significantly (58.4% vs. 31.2%).
Figure 4 further demonstrates OlymMATHâs reliability by comparing OlymMATH-EN performance against AIME24. The close clustering of data points around linear trend lines indicates consistent relative model performance across both benchmarks. This strong correlation suggests OlymMATH measures similar underlying mathematical reasoning abilities as the respected AIME24 dataset, validating its use for LLM evaluation (see Figure 8 in Appendix for more information). Despite this alignment, OlymMATH, particularly the HARD subset, remains significantly more challenging than AIME24 for most models, reinforcing its superior ability to differentiate state-of-the-art capabilities.
### 3.4 Case Study
During our data collection and preliminary experiments, we empirically observed that LLMs sometimes resort to empirical guesses âsuch as heuristics, symmetry assumptions, or even fabricationârather than rigorous reasoning. For instance, o3-mini-high merely âguessedâ $b=c$ due to symmetry in a geometric optimization problem (see Figure 9 in Appendix). While such intuitive approaches might yield correct answers, they lack logical rigor and this case becomes problematic when employing rule-based or LLM-as-judge methods, as neither can effectively assess the quality of rigorous reasoning, thus potentially leading to an illusory improvement in accuracy via âshortcutsâ.
Similar issues were observed in the AIME 2025 and Omni-MATH benchmarks (see Figure 10 and 11 in Appendix), indicating that despite performance gains, LLMs still exhibit deficiencies in deliberative thinking. This underscores the importance of process-level supervision, though its scalability remains a challenge. Currently, we do not accurately measure the proportion of âguessesâ in these benchmarks, leaving this as an important direction for future work.
Notably, these guessing strategies often fail on our OlymMATH dataset. For example, a model incorrectly assumed symmetry for a complex optimization problem in OlymMATH, yielding $3081$ instead of the correct $2625$ (see Figure 12 in Appendix). OlymMATH problems, particularly in the HARD subset, are selected and designed so that their reasoning steps are difficult to âhackâ through empirical guessing, thus providing a more robust evaluation of genuine reasoning capabilities.
## 4 Usability and Accessibility
To support research into LLM reasoning, we have open-sourced the OlymMATH-eval dataset at https://hf.co/datasets/RUC-AIBOX/OlymMATH-eval, with 582,400 entries from 28 models, to help compare reasoning capabilities across different models and mathematical domains.
Furthermore, we provide the OlymMATH-demo visualization tool (https://hf.co/spaces/RUC-AIBOX/OlymMATH-demo; see Figure 5) to facilitate in-depth analysis of LLM reasoning. This interactive interface enables: (1) Side-by-side comparison of two selected LLMs on the same L a T e X -rendered problem, with access to reference answers. (2) Color-coded âProblem Gridsâ for each model, displaying per-problem accuracy for quick identification of challenging areas. (3) Examination of individual model-generated reasoning samples, including correctness, extracted answers, and token counts, crucial for understanding solution processes and identifying flaws. The tool also includes standard solutions for difficult problems and supports local deployment. OlymMATH-demo is thus a valuable asset for dissecting reasoning patterns, diagnosing errors, and guiding LLM development.
<details>
<summary>extracted/6453487/figs/demo.png Details</summary>
### Visual Description
# Technical Document Extraction
## Overview
The image contains two comparative models (Model 1 and Model 2) with problem grids, sample data, and accuracy metrics. Text is primarily in Chinese with English annotations. Key components include heatmaps, sample number inputs, and financial/token metrics.
---
## Model 1
### Problem Grid
- **Structure**: 10x10 matrix (0-99) with percentage values
- **Color Legend**:
- Green: 0-20%
- Yellow: 21-40%
- Orange: 41-60%
- Red: 61-80%
- Dark Red: 81-100%
- **Sample Numbers**: 8 samples (0-63)
- **Accuracy**: 0/8 = 0.0%
- **Extracted Data**:
- Financial: $3360
- Tokens: 16133
### Chinese Text Translation
> "For quadrilateral ABCD - AâBâCâDâ, divide 1,2,...,8 into eight parts on the quadrilateral's vertices. Each face requires three numbers, and the average should not be less than 10. Find different numbers for each."
---
## Model 2
### Problem Grid
- **Structure**: 10x10 matrix (0-99) with percentage values
- **Color Legend**:
- Green: 0-20%
- Yellow: 21-40%
- Orange: 41-60%
- Red: 61-80%
- Dark Red: 81-100%
- **Sample Numbers**: 64 samples (0-63)
- **Accuracy**: 14/64 = 21.9%
- **Extracted Data**:
- Financial: $480
- Tokens: 12751
### Chinese Text Translation
> "Each face has four vertices, but the problem requires three numbers per face. After selecting different numbers, the average should not be less than 10."
---
## Spatial Analysis
1. **Legend Placement**:
- Model 1: Top-left corner
- Model 2: Top-right corner
2. **Color Consistency**:
- Verified all grid cells match legend color ranges
- Example: Model 1 cell 0 (0%) = Green (0-20%)
---
## Trend Verification
- **Model 1 Grid Trends**:
- Highest values (81-100%) concentrated in lower rows (80-99)
- Lower values (0-20%) in upper rows (0-20)
- **Model 2 Grid Trends**:
- More distributed values with 21.9% accuracy indicating moderate performance
---
## Component Isolation
1. **Header**:
- "Problem Statement" (blue) and "Reference Answer" tabs
2. **Main Charts**:
- Two heatmaps with percentage distributions
3. **Footer**:
- Sample number inputs and extracted metrics
---
## Data Table Reconstruction
### Model 1 Sample 8
| Sample | Value | Color | Accuracy |
|--------|-------|--------|----------|
| 0 | 0 | Green | 0% |
| 1 | 1 | Red | 100% |
| ... | ... | ... | ... |
| 7 | 7 | Red | 100% |
### Model 2 Sample 64
| Sample | Value | Color | Accuracy |
|--------|-------|--------|----------|
| 0 | 0 | Green | 100% |
| 1 | 1 | Red | 100% |
| ... | ... | ... | ... |
| 63 | 63 | Red | 100% |
---
## Language Notes
- **Primary Language**: Chinese (Simplified)
- **Secondary Language**: English (annotations)
- **Translated Text**: Provided for critical problem statements
---
## Conclusion
The image compares two models with distinct performance metrics. Model 2 shows significantly better accuracy (21.9% vs 0.0%) despite similar grid structures. Financial and token metrics suggest different computational costs between models.
</details>
Figure 5: The OlymMATH-demo interface. It is currently being maintained on HuggingFace Spaces.
## 5 Conclusion
We introduced OlymMATH, a challenging math benchmark for LLMs, uniquely curated from printed materials. It includes 200 problems across four fields, with easy (AIME-level) and hard (more challenging) subsets, in parallel English and Chinese versions. Our experiments with state-of-the-art LLMs, especially in slow-thinking modes, show significant struggles. Analysis highlights language-specific strengths and universal limitations like empirical guessing, identifying weaknesses in LLMsâ multi-step reasoning and logical consistency. Meanwhile, to facilitate community research, we have open-sourced over 580k reasoning data, a visualization tool, and solutions for challenging problems.
As part of our STILL project, OlymMATH affirms our belief in benchmarksâ pivotal role in advancing LLMsâ reasoning capabilities. We advocate for benchmarks to evolve faster than methodologies, guiding the fieldâs progress. Our planned expansion of OlymMATH embodies this commitment, aiming to further cultivate the development of more robust reasoning models and continue pushing the boundaries of language intelligence.
## References
- [1] Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, Yifan Du, Chen Yang, Yushuo Chen, Zhipeng Chen, Jinhao Jiang, Ruiyang Ren, Yifan Li, Xinyu Tang, Zikang Liu, Peiyu Liu, Jian-Yun Nie, and Ji-Rong Wen. A survey of large language models, 2025.
- [2] An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, Junyang Lin, Kai Dang, Keming Lu, Keqin Bao, Kexin Yang, Le Yu, Mei Li, Mingfeng Xue, Pei Zhang, Qin Zhu, Rui Men, Runji Lin, Tianhao Li, Tingyu Xia, Xingzhang Ren, Xuancheng Ren, Yang Fan, Yang Su, Yichang Zhang, Yu Wan, Yuqiong Liu, Zeyu Cui, Zhenru Zhang, and Zihan Qiu. Qwen2.5 technical report. arXiv preprint arXiv:2412.15115, 2024.
- [3] Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Arthur Hinsvark, Arun Rao, Aston Zhang, Aurélien Rodriguez, Austen Gregerson, Ava Spataru, Baptiste RoziÚre, Bethany Biron, Binh Tang, Bobbie Chern, Charlotte Caucheteux, Chaya Nayak, Chloe Bi, Chris Marra, Chris McConnell, Christian Keller, Christophe Touret, Chunyang Wu, Corinne Wong, Cristian Canton Ferrer, Cyrus Nikolaidis, Damien Allonsius, Daniel Song, Danielle Pintz, Danny Livshits, David Esiobu, Dhruv Choudhary, Dhruv Mahajan, Diego Garcia-Olano, Diego Perino, Dieuwke Hupkes, Egor Lakomkin, Ehab AlBadawy, Elina Lobanova, Emily Dinan, Eric Michael Smith, Filip Radenovic, Frank Zhang, Gabriel Synnaeve, Gabrielle Lee, Georgia Lewis Anderson, Graeme Nail, Grégoire Mialon, Guan Pang, Guillem Cucurell, Hailey Nguyen, Hannah Korevaar, Hu Xu, Hugo Touvron, Iliyan Zarov, Imanol Arrieta Ibarra, Isabel M. Kloumann, Ishan Misra, Ivan Evtimov, Jade Copet, Jaewon Lee, Jan Geffert, Jana Vranes, Jason Park, Jay Mahadeokar, Jeet Shah, Jelmer van der Linde, Jennifer Billock, Jenny Hong, Jenya Lee, Jeremy Fu, Jianfeng Chi, Jianyu Huang, Jiawen Liu, Jie Wang, Jiecao Yu, Joanna Bitton, Joe Spisak, Jongsoo Park, Joseph Rocca, Joshua Johnstun, Joshua Saxe, Junteng Jia, Kalyan Vasuden Alwala, Kartikeya Upasani, Kate Plawiak, Ke Li, Kenneth Heafield, Kevin Stone, and et al. The llama 3 herd of models. CoRR, abs/2407.21783, 2024.
- [4] DeepSeek-AI, Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, Xiaokang Zhang, Xingkai Yu, Yu Wu, Z. F. Wu, Zhibin Gou, Zhihong Shao, Zhuoshu Li, Ziyi Gao, Aixin Liu, Bing Xue, Bingxuan Wang, Bochao Wu, Bei Feng, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenyu Zhang, Chong Ruan, Damai Dai, Deli Chen, Dongjie Ji, Erhang Li, Fangyun Lin, Fucong Dai, Fuli Luo, Guangbo Hao, Guanting Chen, Guowei Li, H. Zhang, Han Bao, Hanwei Xu, Haocheng Wang, Honghui Ding, Huajian Xin, Huazuo Gao, Hui Qu, Hui Li, Jianzhong Guo, Jiashi Li, Jiawei Wang, Jingchang Chen, Jingyang Yuan, Junjie Qiu, Junlong Li, J. L. Cai, Jiaqi Ni, Jian Liang, Jin Chen, Kai Dong, Kai Hu, Kaige Gao, Kang Guan, Kexin Huang, Kuai Yu, Lean Wang, Lecong Zhang, Liang Zhao, Litong Wang, Liyue Zhang, Lei Xu, Leyi Xia, Mingchuan Zhang, Minghua Zhang, Minghui Tang, Meng Li, Miaojun Wang, Mingming Li, Ning Tian, Panpan Huang, Peng Zhang, Qiancheng Wang, Qinyu Chen, Qiushi Du, Ruiqi Ge, Ruisong Zhang, Ruizhe Pan, Runji Wang, R. J. Chen, R. L. Jin, Ruyi Chen, Shanghao Lu, Shangyan Zhou, Shanhuang Chen, Shengfeng Ye, Shiyu Wang, Shuiping Yu, Shunfeng Zhou, Shuting Pan, and S. S. Li. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. CoRR, abs/2501.12948, 2025.
- [5] OpenAI. Openai o1 system card, 2024.
- [6] Zhipeng Chen, Yingqian Min, Beichen Zhang, Jie Chen, Jinhao Jiang, Daixuan Cheng, Wayne Xin Zhao, Zheng Liu, Xu Miao, Yang Lu, Lei Fang, Zhongyuan Wang, and Ji-Rong Wen. An empirical study on eliciting and improving r1-like reasoning models, 2025.
- [7] An Yang, Beichen Zhang, Binyuan Hui, Bofei Gao, Bowen Yu, Chengpeng Li, Dayiheng Liu, Jianhong Tu, Jingren Zhou, Junyang Lin, Keming Lu, Mingfeng Xue, Runji Lin, Tianyu Liu, Xingzhang Ren, and Zhenru Zhang. Qwen2.5-math technical report: Toward mathematical expert model via self-improvement, 2024.
- [8] Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, et al. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168, 2021.
- [9] Dan Hendrycks, Collin Burns, Saurav Kadavath, Akul Arora, Steven Basart, Eric Tang, Dawn Song, and Jacob Steinhardt. Measuring mathematical problem solving with the MATH dataset. In Proceedings of the Neural Information Processing Systems Track on Datasets and Benchmarks 1, NeurIPS Datasets and Benchmarks 2021, December 2021, virtual, 2021.
- [10] Meng Fang, Xiangpeng Wan, Fei Lu, Fei Xing, and Kai Zou. Mathodyssey: Benchmarking mathematical problem-solving skills in large language models using odyssey math data, 2024.
- [11] Daman Arora, Himanshu Gaurav Singh, and Mausam. Have llms advanced enough? a challenging problem solving benchmark for large language models, 2023.
- [12] OpenAI. Openai o3-mini: Pushing the frontier of cost-effective reasoning, 1 2025.
- [13] Google Deepmind. Gemini 2.5: Our most intelligent ai model, 3 2025.
- [14] Yingqian Min, Zhipeng Chen, Jinhao Jiang, Jie Chen, Jia Deng, Yiwen Hu, Yiru Tang, Jiapeng Wang, Xiaoxue Cheng, Huatong Song, Wayne Xin Zhao, Zheng Liu, Zhongyuan Wang, and Ji-Rong Wen. Imitate, explore, and self-improve: A reproduction report on slow-thinking reasoning systems, 2024.
- [15] Mathematical Association of America. Aime 2024, 2024.
- [16] Mathematical Association of America. Aime 2025, 2025.
- [17] HMMT. Hmmt 202502, 2025.
- [18] Mathematical Association of America. Usamo 2025, 2025.
- [19] Chaoqun He, Renjie Luo, Yuzhuo Bai, Shengding Hu, Zhen Leng Thai, Junhao Shen, Jinyi Hu, Xu Han, Yujie Huang, Yuxiang Zhang, Jie Liu, Lei Qi, Zhiyuan Liu, and Maosong Sun. Olympiadbench: A challenging benchmark for promoting AGI with olympiad-level bilingual multimodal scientific problems. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), ACL 2024, Bangkok, Thailand, August 11-16, 2024, pages 3828â3850, 2024.
- [20] Bofei Gao, Feifan Song, Zhe Yang, Zefan Cai, Yibo Miao, Qingxiu Dong, Lei Li, Chenghao Ma, Liang Chen, Runxin Xu, et al. Omni-math: A universal olympiad level mathematic benchmark for large language models. arXiv preprint arXiv:2410.07985, 2024.
- [21] RUCAIBox STILL Team. Still-3-1.5b-preview: Enhancing slow thinking abilities of small models through reinforcement learning. 2025.
- [22] Michael Luo, Sijun Tan, Justin Wong, Xiaoxiang Shi, William Y. Tang, Manan Roongta, Colin Cai, Jeffrey Luo, Li Erran Li, Raluca Ada Popa, and Ion Stoica. Deepscaler: Surpassing o1-preview with a 1.5b model by scaling rl, 2025. Notion Blog.
- [23] Qwen Team. Qwq-32b: Embracing the power of reinforcement learning, March 2025.
- [24] Liang Wen, Yunke Cai, Fenrui Xiao, Xin He, Qi An, Zhenyu Duan, Yimin Du, Junchen Liu, Lifu Tang, Xiaowei Lv, Haosheng Zou, Yongchao Deng, Shousheng Jia, and Xiangzheng Zhang. Light-r1: Curriculum sft, dpo and rl for long cot from scratch and beyond, 2025.
- [25] OpenThoughts Team. Open Thoughts. https://open-thoughts.ai, January 2025.
- [26] Jujie He, Jiacai Liu, Chris Yuhao Liu, Rui Yan, Chaojie Wang, Peng Cheng, Xiaoyu Zhang, Fuxiang Zhang, Jiacheng Xu, Wei Shen, Siyuan Li, Liang Zeng, Tianwen Wei, Cheng Cheng, Yang Liu, and Yahui Zhou. Skywork open reasoner series, 2025. Notion Blog.
- [27] Team GLM, Aohan Zeng, Bin Xu, Bowen Wang, Chenhui Zhang, Da Yin, Diego Rojas, Guanyu Feng, Hanlin Zhao, Hanyu Lai, Hao Yu, Hongning Wang, Jiadai Sun, Jiajie Zhang, Jiale Cheng, Jiayi Gui, Jie Tang, Jing Zhang, Juanzi Li, Lei Zhao, Lindong Wu, Lucen Zhong, Mingdao Liu, Minlie Huang, Peng Zhang, Qinkai Zheng, Rui Lu, Shuaiqi Duan, Shudan Zhang, Shulin Cao, Shuxun Yang, Weng Lam Tam, Wenyi Zhao, Xiao Liu, Xiao Xia, Xiaohan Zhang, Xiaotao Gu, Xin Lv, Xinghan Liu, Xinyi Liu, Xinyue Yang, Xixuan Song, Xunkai Zhang, Yifan An, Yifan Xu, Yilin Niu, Yuantao Yang, Yueyan Li, Yushi Bai, Yuxiao Dong, Zehan Qi, Zhaoyu Wang, Zhen Yang, Zhengxiao Du, Zhenyu Hou, and Zihan Wang. Chatglm: A family of large language models from glm-130b to glm-4 all tools, 2024.
- [28] Zihan Liu, Yang Chen, Mohammad Shoeybi, Bryan Catanzaro, and Wei Ping. Acemath: Advancing frontier math reasoning with post-training and reward modeling. arXiv preprint, 2024.
- [29] Ivan Moshkov, Darragh Hanley, Ivan Sorokin, Shubham Toshniwal, Christof Henkel, Benedikt Schifferer, Wei Du, and Igor Gitman. Aimo-2 winning solution: Building state-of-the-art mathematical reasoning models with openmathreasoning dataset, 2025.
- [30] Qwen Team. Qwen3, April 2025.
## Appendix A Appendix
This part presents the detailed content of the dataset and the case study examples mentioned before.
| Problem: Given that two vertices of an equilateral triangle are on the parabola $y^{2}=4x$ , and the third vertex is on the directrix of the parabola, and the distance from the center of the triangle to the directrix equals $\frac{1}{9}$ of the perimeter. Find the area of the triangle. Subject: Geometry |
| --- |
Figure 6: A geometry problem described precisely in text from OlymMATH.
| Original problem: If the distances from the eight vertices of a cube to a certain plane are $0 0$ , $1$ , $2$ , $3$ , $4$ , $5$ , $6$ , $7$ respectively, what is the possible edge length of this cube? After transformation: If the distances from the eight vertices of a cube to a certain plane are $0 0$ , $1$ , $2$ , $3$ , $4$ , $5$ , $6$ , $7$ respectively, consider all possible edge lengths of this cube. Assuming the possible edge lengths form a set $S$ , find the sum of squares of all elements in $S$ . |
| --- |
Figure 7: An OlymMATH-HARD example testing modelâs identification of all possible answers.
<details>
<summary>x4.png Details</summary>
### Visual Description
# Technical Document Extraction: Box Plot Analysis of Accuracy Across Datasets
## Figure Caption
**Figure 1**: Box plots comparing accuracy distributions across five datasets: AIME24, EN-EASY, EN-HARD, ZH-EASY, and ZH-HARD. Accuracy values range from 0.0 to 1.0.
---
## Key Components and Spatial Grounding
1. **Axes**:
- **Y-Axis**: Labeled "Accuracy" with a linear scale from 0.0 to 1.0 (increments of 0.2).
- **X-Axis**: Categorical axis listing datasets: AIME24, EN-EASY, EN-HARD, ZH-EASY, ZH-HARD.
- **Grid**: Dashed gray grid lines for reference.
2. **Legend**:
- No explicit legend is present in the image. Colors are directly mapped to datasets as follows:
- **Purple**: AIME24
- **Blue**: EN-EASY
- **Red**: EN-HARD
- **Green**: ZH-EASY
- **Orange**: ZH-HARD
3. **Outliers**:
- Represented by black circles (â) outside the whiskers of each box plot.
---
## Dataset-Specific Analysis
### AIME24 (Purple)
- **Median Accuracy**: ~0.75 (horizontal line within the box).
- **Interquartile Range (IQR)**: ~0.65â0.80 (box height).
- **Outliers**: Two points at ~0.30 and ~0.35 (below the lower whisker).
### EN-EASY (Blue)
- **Median Accuracy**: ~0.65.
- **IQR**: ~0.55â0.70.
- **Outliers**: Two points at ~0.20 and ~0.25 (below the lower whisker).
### EN-HARD (Red)
- **Median Accuracy**: ~0.15.
- **IQR**: ~0.10â0.20.
- **Outliers**: One point at ~0.35 (above the upper whisker).
### ZH-EASY (Green)
- **Median Accuracy**: ~0.60.
- **IQR**: ~0.40â0.70.
- **Outliers**: One point at ~0.55 (above the upper whisker).
### ZH-HARD (Orange)
- **Median Accuracy**: ~0.10.
- **IQR**: ~0.05â0.15.
- **Outliers**: One point at ~0.30 (above the upper whisker).
---
## Key Trends and Observations
1. **Highest Accuracy**:
- **AIME24** demonstrates the highest median accuracy (~0.75) and the widest IQR (~0.65â0.80), indicating robust performance with moderate variability.
2. **Lowest Accuracy**:
- **EN-HARD** and **ZH-HARD** exhibit the lowest medians (~0.15 and ~0.10, respectively), with narrow IQRs, suggesting poor and consistent performance.
3. **Intermediate Performance**:
- **EN-EASY** and **ZH-EASY** show mid-range accuracy (~0.65 and ~0.60 medians). EN-EASY has a wider IQR (~0.55â0.70), indicating higher variability compared to ZH-EASY (~0.40â0.70).
4. **Outlier Patterns**:
- Outliers are sparse and concentrated below the lower whiskers for AIME24, EN-EASY, and ZH-HARD. EN-HARD and ZH-EASY have outliers above their upper whiskers.
---
## Data Table Reconstruction (Summary)
| Dataset | Median Accuracy | Lower Quartile | Upper Quartile | Outlier(s) |
|-----------|-----------------|----------------|----------------|---------------------|
| AIME24 | 0.75 | 0.65 | 0.80 | 0.30, 0.35 |
| EN-EASY | 0.65 | 0.55 | 0.70 | 0.20, 0.25 |
| EN-HARD | 0.15 | 0.10 | 0.20 | 0.35 |
| ZH-EASY | 0.60 | 0.40 | 0.70 | 0.55 |
| ZH-HARD | 0.10 | 0.05 | 0.15 | 0.30 |
---
## Color Consistency Verification
- All elements (box, median line, whiskers) for each dataset match the assigned color:
- AIME24: Purple (box, median, whiskers).
- EN-EASY: Blue (box, median, whiskers).
- EN-HARD: Red (box, median, whiskers).
- ZH-EASY: Green (box, median, whiskers).
- ZH-HARD: Orange (box, median, whiskers).
- Outliers (black circles) are consistent across all datasets.
---
## Conclusion
The box plots reveal significant disparities in accuracy across datasets. AIME24 and ZH-EASY outperform EN-HARD and ZH-HARD, while EN-EASY and ZH-EASY show moderate performance. Outliers suggest occasional deviations from typical accuracy ranges, particularly in harder datasets (EN-HARD, ZH-HARD).
</details>
Figure 8: This boxplot shows that our EASY dataset has AIME-level difficulty with a wider distribution, while the HARD dataset is significantly more challenging, demonstrating the discriminative power of our benchmark.
| Problem: In the triangle $\triangle ABC$ , the lengths of the three sides are $a,b,c$ , satisfying the condition $2a^{2}+b^{2}+c^{2}=4$ . Find the maximum value of the area of $\triangle ABC$ . Answer: $\frac{\sqrt{5}}{5}$ . (When $a=\frac{2}{\sqrt{5}}$ , $b=c=\sqrt{\frac{6}{5}}$ ) |
| --- |
| Model Output: We wish to maximize the area of $\triangle ABC$ , subject to $2a^{2}+b^{2}+c^{2}=4.$ Because the condition is symmetric in $b$ and $c$ (but not in $a$ ), it is reasonable to look for an optimum among triangles having $b=c$ . (Later one can check that assuming $b=c$ indeed gives the maximum.) The problem involves maximizing the area of an isosceles triangle with sides $a,b,b$ under the constraint $2a^{2}+b^{2}+c^{2}=4$ . By assuming $b=c$ , we derive a formula for the area in terms of $b$ and find the maximum area by expressing it as a function of $u=b^{2}$ . We then determine the critical points of this quadratic function, identify the maximum area, and finally recover the side lengths to verify that they satisfy the original constraints. The maximum area achieved is $\frac{\sqrt{5}}{5}$ . |
Figure 9: An example during our data collection. o3-mini (high) found the correct answer without rigorously solving the problem. Due to length constraints, we provide only a summary of its original solution steps.
| Col1 |
| --- |
| Problem: Let the sequence of rationals $x_{1},x_{2},\dots$ be defined such that $x_{1}=\frac{25}{11}$ and $x_{k+1}=\frac{1}{3}\left(x_{k}+\frac{1}{x_{k}}-1\right).$ $x_{2025}$ can be expressed as $\frac{m}{n}$ for relatively prime positive integers $m$ and $n$ . Find the remainder when $m+n$ is divided by $1000$ . Answer: $248$ . |
| Model Output: We will show that if $x_{1}=\frac{25}{11}$ and $x_{k+1}=\frac{1}{3}\Bigl{(}x_{k}+\frac{1}{x_{k}}-1\Bigr{)},$ then after âmanyâ iterations the answer to $x_{2025}=\frac{m}{n}$ (in lowest terms) satisfies $m+n\equiv 248\pmod{1000}.$ In what follows we briefly describe one way (via a âmiraculousâ change-ofâvariables) to âsolveâ the recurrence; the details need not be reproduced fully here. A good hint is that the recurrence $x_{k+1}=\frac{1}{3}\Bigl{(}x_{k}+\frac{1}{x_{k}}-1\Bigr{)}$ has the curious symmetry $T(x)=\frac{1}{3}\Bigl{(}x+\frac{1}{x}-1\Bigr{)}=T\Bigl{(}\frac{1}{x}\Bigr{)},$ so the expression $x+\frac{1}{x}$ (which is âsymmetricâ in $x$ and its reciprocal) naturally appears. In fact, one may rewrite the recurrence in the form $3x_{k+1}+1=x_{k}+\frac{1}{x_{k}}.$ A useful idea is then to âcomplete the squareâ by writing $\left(\sqrt{x_{k}}-\frac{1}{\sqrt{x_{k}}}\right)^{2}=x_{k}+\frac{1}{x_{k}}-2$ , so that $3x_{k+1}+1=\left(\sqrt{x_{k}}-\frac{1}{\sqrt{x_{k}}}\right)^{2}+2.$ Defining $u_{k}=\sqrt{x_{k}}-\frac{1}{\sqrt{x_{k}}}$ , one obtains the âlinear-lookingâ relation $u_{k}^{2}=3x_{k+1}-1.$ With some additional work it turns out that if we write $x_{k}=\frac{m_{k}}{n_{k}}$ (in lowest terms), then one may prove by induction that $m_{k}+n_{k}=36\cdot 12^{(2^{\,k-1}-1)}.$ (In our case one checks that $m_{1}+n_{1}=25+11=36$ , $m_{2}+n_{2}=432$ , $m_{3}+n_{3}=62208,$ and indeed $36,432,62208$ equal $36\cdot 12^{0},\,36\cdot 12^{1},\,36\cdot 12^{3}$ respectively.) Thus, for $k=2025$ we have $m+n=36\cdot 12^{\left(2^{2024}-1\right)}.$ Even though this number is astronomically large, we are only asked to compute $m+n\pmod{1000}.$ Using modular reductions, we compute $m+n$ modulo 1000. First, modulo 8, we find $36\cdot 12^{(2^{2024}-1)}\equiv 0\pmod{8}$ . Modulo 125, Eulerâs theorem helps reduce $12^{N}$ to $12^{15}\equiv 118\pmod{125}$ , giving $36\cdot 12^{N}\equiv 123\pmod{125}$ . Using the Chinese Remainder Theorem, solving $125k+123\equiv 0\pmod{8}$ gives $k\equiv 1\pmod{8}$ , leading to the final answer: $\boxed{248}$ . |
Figure 10: An example from AIME 2025. o3-mini (high) forgot to prove that $(m_{k},n_{k})=1$ and it claimed that $m_{k}+n_{k}=36\cdot 12^{\left(2^{k-1}-1\right)}$ without providing the complete proof (highlighted in red). Instead, the model âguessedâ this expression by calculating the first several terms in the sequence one by one. Due to length constraints, we provide only a summary of its unimportant steps.
Figure 11: An example from Omni-MATH. The solution provided by Omni-MATH itself is flawed and problematic. Both the reference solution and o3-mini (high) failed to prove that there are no other possible function $f$ that satisfies the condition (highlighted in red). Due to length constraints, we have omitted its unimportant steps.
Figure 12: An example from OlymMATH-EN-HARD subset. o3-mini (high) attempted to âguessâ the equality conditions but failed to produce a correct solution (highlighted in red).