Difficulty Correlation with LLM Performance on Omni-Math
0.91Pearson PCCLLM compare
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| LLM compareModel=OpenAI o32025.12 | 0.91 | 0.9 | 0.77 | |
| LLM compareModel=Gemini 2.5 Pro2025.12 | 0.9 | 0.89 | 0.76 | |
| LLM compareModel=OpenAI o32025.12 | 0.24 | 0.25 | 0.21 | |
| LLM compareModel=Gemini 2.5 Pro2025.12 | 0.22 | 0.23 | 0.18 |