Mathematical Reasoning on MATH (Accuracy Improvement Delta)
10.2Accuracy Improvement DeltaUltra-Dense Prompting
Evaluation results
| Method | Links | |
|---|---|---|
| Ultra-Dense PromptingModel=Gemini 2.02026.04 | 10.2 | |
| Ultra-Dense PromptingModel=Claude 3.72026.04 | 9.4 | |
| Ultra-Dense PromptingModel=Average across Models2026.04 | 9.1 | |
| Ultra-Dense PromptingModel=GPT-4o2026.04 | 8.8 | |
| Ultra-Dense PromptingModel=GPT-4o-mini2026.04 | 8.1 |